From matt at linuxbox.com Sat Dec 6 11:06:20 2008 From: matt at linuxbox.com (Matt Benjamin) Date: Sat Dec 6 11:06:27 2008 Subject: OpenAFS on FreeBSD Progress (Works) Message-ID: <493ACAC4.5020806@linuxbox.com> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 Hi Jason (and the rest of this apparently small list), I apologize for not piping up before, I didn't join freebsd-afs when I was copied on an initial flurry of emails, but just did. I have the following progress: 1. I finished initial work fixing open issues with the stalled FreeBSD client cache manager in OpenAFs, and made it run on i386 FreeBSD 7.0 and 8-Current [as of around Sept. 15 or so]. 2. The i386 FreeBSD 7.0 target is now an "officially supported platform" in OpenAFS 1.4.8--though the client has only recently been finished, and probably has undiscovered issues--the only issues I'm actually aware of for certain are a cosmetic, and I use the port myself 3. This week I posted a patch to OpenAFS RT (our bug tracker) adding x86_64 FreeBSD 7.0 support, with some help from Tony Jago. I'm expecting to see the x86_64 platform officially supported in the next stable OpenAFS release Just as extra information, I've also worked on the i386 OpenBSD port, updating it for OpenBSD 4.4, and am most of the way through a new port to NetBSD 4.0, based on the existing OpenBSD port (which was itself based on an ancient NetBSD port, in fact). I would appreciate any help and guidance getting OpenAFS more fully usable in FreeBSD. Matt OpenAFS on FreeBSD Progress Jason C. Wells jcw at highperformance.net Mon Sep 22 03:57:56 UTC 2008 * Previous message: AFS Hackathon - October 27-28 at Google HQ, Mountain View, CA, USA * Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] At the beginning of the year, there was a discussion on a port of OpenAFS, especially the client/cache manager for FreeBSD. Did this discussion bear any fruit? I'm considering switching to Debian on my desktop to get OpenAFS. I'd wait if I knew an OpenAFS port was forthcoming. Regards, Jason - -- Matt Benjamin The Linux Box 206 South Fifth Ave. Suite 150 Ann Arbor, MI 48104 http://linuxbox.com tel. 734-761-4689 fax. 734-769-8938 cel. 734-216-5309 -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.7 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFJOsrEJiSUUSaRdSURCI9UAJ455pRtBv79aVN6FzNZuHiIStJ1mwCfV18i 4OFN+p5UmRmcYuIGoDiaiBs= =f62N -----END PGP SIGNATURE----- From jcw at highperformance.net Sat Dec 6 11:38:20 2008 From: jcw at highperformance.net (Jason C. Wells) Date: Sat Dec 6 11:38:26 2008 Subject: OpenAFS on FreeBSD Progress (Works) In-Reply-To: <493ACAC4.5020806@linuxbox.com> References: <493ACAC4.5020806@linuxbox.com> Message-ID: <493ACBFA.20909@highperformance.net> Great news Matt! I can't wait to try it. I have a big upgrade planned in the near future and will work in AFS client support at the same time. Later, Jason From bsam at ipt.ru Mon Dec 8 12:49:17 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Mon Dec 8 12:49:24 2008 Subject: OpenAFS on FreeBSD Progress (Works) In-Reply-To: <493ACAC4.5020806@linuxbox.com> (Matt Benjamin's message of "Sat\, 06 Dec 2008 13\:56\:04 -0500") References: <493ACAC4.5020806@linuxbox.com> Message-ID: <12501719@bb.ipt.ru> Hello List, Matt Benjamin writes: > Hi Jason (and the rest of this apparently small list), > > I apologize for not piping up before, I didn't join freebsd-afs when I > was copied on an initial flurry of emails, but just did. > > I have the following progress: Good work, thank you and all involved very much! > 1. I finished initial work fixing open issues with the stalled FreeBSD > client cache manager in OpenAFs, and made it run on i386 FreeBSD 7.0 and > 8-Current [as of around Sept. 15 or so]. > > 2. The i386 FreeBSD 7.0 target is now an "officially supported platform" > in OpenAFS 1.4.8--though the client has only recently been finished, and > probably has undiscovered issues--the only issues I'm actually aware of > for certain are a cosmetic, and I use the port myself > > 3. This week I posted a patch to OpenAFS RT (our bug tracker) adding > x86_64 FreeBSD 7.0 support, with some help from Tony Jago. I'm > expecting to see the x86_64 platform officially supported in the next > stable OpenAFS release Well, I've just finish a FreeBSD port. It is based on port by Alec Kloss (CCed0 which was sent to me by Tony Jago (CCed) by email. Actually I wrote some comments about the port I've got but haven't received a reply so far. Let me introduce the port: ftp://ftp.ipt.ru/pub/tmp/openafs.shar or one can browse it here: ftp://ftp.ipt.ru/pub/tmp/openafs . The port was tested at tinderbox for 7.0-i386, 7-STABLE-i386, 8-CURRENT-i386, 7-STABLE-amd64, 8-CURRENT-amd64. The port uses sources from 1.4.8 with integrated patches for amd64. Actually, there are two extra patches. the first one is needed for building at amd64. It should be committed along with those from OpenAFS RT (ftp://ftp.ipt.ru/pub/tmp/openafs/files/patch-configure). While the second one is a real hack to make openafs to compile at 8-CURRENT (ftp://ftp.ipt.ru/pub/tmp/openafs/files/patch-curseswindows.c). I'd like to create two separate ports: openafs-server and openafs-client. But I don't know right now how to do it. Any comments are appreciated. WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From matt at linuxbox.com Mon Dec 8 12:54:48 2008 From: matt at linuxbox.com (Matt Benjamin) Date: Mon Dec 8 12:54:54 2008 Subject: OpenAFS on FreeBSD Progress (Works) In-Reply-To: <12501719@bb.ipt.ru> References: <493ACAC4.5020806@linuxbox.com> <12501719@bb.ipt.ru> Message-ID: <493D898C.1030609@linuxbox.com> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 Hi Boris (CC others), Thanks for your work, in turn! As you can imagine, since I'm from the OpenAFS side, I have the following agenda: 1. get official FreeBSD port of openafs server and client established--it sounds like you're on track to do this (more below) 2. absorb all relevant patches into OpenAFS upstream repo--we have Tony's 64bit fixes, as you saw, and have built these into what I submitted to RT for AMD64 target; thanks for the pointer to the other two patches, and for putting them into RT already (!); I'll look at those shortly Note that I have submitted additional patches for FreeBSD openafs server and client that will probably appear in the next stable release--including masking SIGSYS triggered by ktc_GetToken as used by vos and other programs, when the lkm is not loaded [ugh], sorry didn't fix that earlier. Now, returning to the port topic... - From what I understand, Alec's port uses Arla as the client, so I think that port isn't the _client_ port I personally and OpenAFS are trying for, but presumably they can coexist? Alec's is ideal for sites wishing to blend OpenAFS with Arla as the client, &c, and we certainly want to assist in any way we can with that effort, as well. I would like to open discussion on how OpenAFS specifically should best interface with the FreeBSD ports system--and people who would like to help us keep that port up to date :) Regards, Matt Boris Samorodov wrote: > > I'd like to create two separate ports: openafs-server and > openafs-client. But I don't know right now how to do it. > > Any comments are appreciated. > > > WBR - -- Matt Benjamin The Linux Box 206 South Fifth Ave. Suite 150 Ann Arbor, MI 48104 http://linuxbox.com tel. 734-761-4689 fax. 734-769-8938 cel. 734-216-5309 -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.7 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFJPYmMJiSUUSaRdSURCAKrAJ0SZfZlY8UYIas7h5tshzUYn8iwLACeNPzK CCbpepY7BVZb8SYKrF5nzlY= =OY6z -----END PGP SIGNATURE----- From afs at spam.t71.org Mon Dec 8 16:11:48 2008 From: afs at spam.t71.org (Tony Jago) Date: Mon Dec 8 16:11:55 2008 Subject: OpenAFS on FreeBSD Progress (Works) In-Reply-To: <493D898C.1030609@linuxbox.com> References: <493ACAC4.5020806@linuxbox.com> <12501719@bb.ipt.ru> <493D898C.1030609@linuxbox.com> Message-ID: <22B6C509EF7C4AB0A2D8350C31BB8D5D@valentine> > Note that I have submitted additional patches for FreeBSD openafs server > and client that will probably appear in the next stable > release--including masking SIGSYS triggered by ktc_GetToken as used by > vos and other programs, when the lkm is not loaded [ugh], sorry didn't > fix that earlier. That would be excellent. Another option would be to have many of the commands default to -localauth when the kernel module isn't loaded as this is probably what the user was wanting anyway. On this topic, it's impossible to setup an openafs server without a working afs client as you need to setup the root.cell and set permissions. For this reason I think afs will become a lot more used when the client is working. >From my personal testing, the server 1.4.8 on freebsd (on AMD and i386) work really well. We have being using it in light production mode for some time in freebsd jails. The client I have had problems with. When testing I found it corrupts files. It can't handle symlinks and is just a little rusty at the moment. I haven't tried the 1.4.8 client but will give it a go shortly to see how things go. > - From what I understand, Alec's port uses Arla as the client, so I think > that port isn't the _client_ port I personally and OpenAFS are trying > for, but presumably they can coexist? Alec's is ideal for sites wishing > to blend OpenAFS with Arla as the client, &c, and we certainly want to > assist in any way we can with that effort, as well. Both ports can co-exist in the freebsd ports system. They probably should be marked to exclude each other. > I would like to open discussion on how OpenAFS specifically should best > interface with the FreeBSD ports system--and people who would like to > help us keep that port up to date :) >> I'd like to create two separate ports: openafs-server and >> openafs-client. But I don't know right now how to do it. I would also recommend a openafs metaport that combines the client and the server. Have a look at the mysql-client, mysql-server and mysql ports to see an example of how to do this sort of thing. I am really looking forward to a stable afs on freebsd. It's been a long wait. This combined with a ZFS filesystem in the backend is a very very powerful system. Tony From shadow at gmail.com Mon Dec 8 16:43:26 2008 From: shadow at gmail.com (Derrick Brashear) Date: Mon Dec 8 16:43:32 2008 Subject: OpenAFS on FreeBSD Progress (Works) In-Reply-To: <22B6C509EF7C4AB0A2D8350C31BB8D5D@valentine> References: <493ACAC4.5020806@linuxbox.com> <12501719@bb.ipt.ru> <493D898C.1030609@linuxbox.com> <22B6C509EF7C4AB0A2D8350C31BB8D5D@valentine> Message-ID: On Mon, Dec 8, 2008 at 6:55 PM, Tony Jago wrote: > Note that I have submitted additional patches for FreeBSD openafs server >> and client that will probably appear in the next stable >> release--including masking SIGSYS triggered by ktc_GetToken as used by >> vos and other programs, when the lkm is not loaded [ugh], sorry didn't >> fix that earlier. >> > > That would be excellent. Another option would be to have many of the > commands default to -localauth when the kernel module isn't loaded as this > is probably what the user was wanting anyway. Unless they're horribly broken they will be there. > > On this topic, it's impossible to setup an openafs server without a working > afs client as you need to setup the root.cell and set permissions. For this > reason I think afs will become a lot more used when the client is working. > Annoying, maybe. Hardly impossible. -- Derrick From matt at linuxbox.com Mon Dec 8 16:44:30 2008 From: matt at linuxbox.com (Matt Benjamin) Date: Mon Dec 8 16:44:37 2008 Subject: OpenAFS on FreeBSD Progress (Works) In-Reply-To: <22B6C509EF7C4AB0A2D8350C31BB8D5D@valentine> References: <493ACAC4.5020806@linuxbox.com> <12501719@bb.ipt.ru> <493D898C.1030609@linuxbox.com> <22B6C509EF7C4AB0A2D8350C31BB8D5D@valentine> Message-ID: <493DBF65.9010402@linuxbox.com> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 Hi Tony, 1. I can't reproduce any problem with symlinks, and I've tested the obvious things--can you put a bug in openafs RT with details? 2. File data consistency. That is noted, please put it in RT as well, if you could--the simpler the steps to reproduce (less complex i/o, iozone, and other dependencies, the easier to fix, likely)--there haven't been many testers yet, but we're committed to supporting the client. Thanks! Matt Tony Jago wrote: > The client I have had problems with. When > testing I found it corrupts files. It can't handle symlinks and is just > a little rusty at the moment. I haven't tried the 1.4.8 client but will > give it a go shortly to see how things go. - -- Matt Benjamin The Linux Box 206 South Fifth Ave. Suite 150 Ann Arbor, MI 48104 http://linuxbox.com tel. 734-761-4689 fax. 734-769-8938 cel. 734-216-5309 -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.7 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFJPb9lJiSUUSaRdSURCKMlAJ4xybGtTSmL1yyjAJatb4DCaDi9bACfaH2x 0vSCDZ1OPQrq21jfJUKhwc0= =Vj3u -----END PGP SIGNATURE----- From bsam at ipt.ru Tue Dec 9 09:57:38 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Tue Dec 9 09:57:45 2008 Subject: OpenAFS on FreeBSD Progress (Works) In-Reply-To: <493D898C.1030609@linuxbox.com> (Matt Benjamin's message of "Mon\, 08 Dec 2008 15\:54\:36 -0500") References: <493ACAC4.5020806@linuxbox.com> <12501719@bb.ipt.ru> <493D898C.1030609@linuxbox.com> Message-ID: <39213648@bb.ipt.ru> Hi Matt and others, Matt Benjamin writes: > As you can imagine, since I'm from the OpenAFS side, I have the > following agenda: > > 1. get official FreeBSD port of openafs server and client > established--it sounds like you're on track to do this (more below) Yes. > 2. absorb all relevant patches into OpenAFS upstream repo--we have > Tony's 64bit fixes, as you saw, and have built these into what I > submitted to RT for AMD64 target; thanks for the pointer to the other > two patches, and for putting them into RT already (!); I'll look at > those shortly Seems I wasn't clear enough. I created a port using *your* diff from RT. But added two extra patches. The first one to enable the port to be compiled at amd64 and the second one to enable to compile the port at FreeBSD-CURRENT. Actually, I tried but didn't find a way to submit a followup to your ticket at RT database. > Note that I have submitted additional patches for FreeBSD openafs server > and client that will probably appear in the next stable > release--including masking SIGSYS triggered by ktc_GetToken as used by > vos and other programs, when the lkm is not loaded [ugh], sorry didn't > fix that earlier. > > Now, returning to the port topic... > > From what I understand, Alec's port uses Arla as the client, so I think > that port isn't the _client_ port I personally and OpenAFS are trying Yes, it's a server port. > for, but presumably they can coexist? Alec's is ideal for sites wishing > to blend OpenAFS with Arla as the client, &c, and we certainly want to > assist in any way we can with that effort, as well. > > I would like to open discussion on how OpenAFS specifically should best > interface with the FreeBSD ports system--and people who would like to > help us keep that port up to date :) Seconded. WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From matt at linuxbox.com Tue Dec 9 10:07:48 2008 From: matt at linuxbox.com (Matt Benjamin) Date: Tue Dec 9 10:07:55 2008 Subject: OpenAFS on FreeBSD Progress (Works) In-Reply-To: <39213648@bb.ipt.ru> References: <493ACAC4.5020806@linuxbox.com> <12501719@bb.ipt.ru> <493D898C.1030609@linuxbox.com> <39213648@bb.ipt.ru> Message-ID: <493EB3EB.7000300@linuxbox.com> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 Boris Samorodov wrote: > Hi Matt and others, > > > Matt Benjamin writes: > >> As you can imagine, since I'm from the OpenAFS side, I have the >> following agenda: >> >> 1. get official FreeBSD port of openafs server and client >> established--it sounds like you're on track to do this (more below) > > Yes. > >> 2. absorb all relevant patches into OpenAFS upstream repo--we have >> Tony's 64bit fixes, as you saw, and have built these into what I >> submitted to RT for AMD64 target; thanks for the pointer to the other >> two patches, and for putting them into RT already (!); I'll look at >> those shortly > > Seems I wasn't clear enough. I created a port using *your* diff from > RT. But added two extra patches. The first one to enable the port to > be compiled at amd64 and the second one to enable to compile the port > at FreeBSD-CURRENT. Sorry, I did mistake you. Which patches from RT (#s) do you have? For now, please attach any patch that we don't have. I have OpenAFS compiling at amd64, in #123811, and I thought it compiled at CURRENT (but it maybe stopped at some point), and will merge any reasonable change you have. We need to fix RT access. We really will. Thanks, Matt - -- Matt Benjamin The Linux Box 206 South Fifth Ave. Suite 150 Ann Arbor, MI 48104 http://linuxbox.com tel. 734-761-4689 fax. 734-769-8938 cel. 734-216-5309 -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.7 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFJPrPqJiSUUSaRdSURCJyzAJwPnsh+Emm01KhzfOsaMNxfP3OYqQCfZXJP z9I4E4VJ/Z6aizTCRudtBVk= =om2C -----END PGP SIGNATURE----- From bsam at ipt.ru Tue Dec 9 13:07:49 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Tue Dec 9 13:07:56 2008 Subject: OpenAFS on FreeBSD Progress (Works) In-Reply-To: <493EB3EB.7000300@linuxbox.com> (Matt Benjamin's message of "Tue\, 09 Dec 2008 13\:07\:39 -0500") References: <493ACAC4.5020806@linuxbox.com> <12501719@bb.ipt.ru> <493D898C.1030609@linuxbox.com> <39213648@bb.ipt.ru> <493EB3EB.7000300@linuxbox.com> Message-ID: <24975806@bs1.sp34.ru> Hi Matt, Matt Benjamin writes: > Which patches from RT (#s) do you have? For now, please attach any > patch that we don't have. I have OpenAFS compiling at amd64, in > #123811, I took the diff from this ticket and create patches based on it. > and I thought it compiled at CURRENT (but it maybe stopped at > some point), I test ports with ports-mgmt/tinderbox. This program creates a vanilla environment for building/packaging. The environment of the main system don't influence testing. > and will merge any reasonable change you have. The first one is just like a diff for src/cf/osconf.m4: ftp://ftp.ipt.ru/pub/tmp/openafs/files/patch-configure Without this patch I got an error at amd64 build: ----- ... cd picobj && cc -I. -I. -I/work/a/ports/net/openafs/work/openafs-1.4.8/src/config -I/work/a/ports/net/openafs/work/openafs-1.4.8/include -O2 -pipe -c glue.c + cc -shared -o libafssetpag.so.1.0 picobj/setpag.o picobj/glue.o syscall.o /usr/bin/ld: picobj/setpag.o: relocation R_X86_64_32 can not be used when making a shared object; recompile with -fPIC picobj/setpag.o: could not read symbols: Bad value *** Error code 1 ----- The second patch (I called it a hack) deals with the absence of sgtty.h since __FreeBSD_version=800039: ftp://ftp.ipt.ru/pub/tmp/openafs/files/patch-curseswindows.c A better way may be to use ifdefs here. Without this patch I get an error at 8-CURRENT build: ----- ... cd src && cd gtx && make all ... cc -O -I/work/a/ports/net/openafs/work/openafs-1.4.8/src/config -I. -I. -I/work/a/ports/net/openafs/work/openafs-1.4.8/include -I/work/a/ports/net/openafs/work/openafs-1.4.8/include/afs -I/work/a/ports/net/openafs/work/openafs-1.4.8/include/rx -I/work/a/ports/net/openafs/work/openafs-1.4.8 -I/work/a/ports/net/openafs/work/openafs-1.4.8/src -I/work/a/ports/net/openafs/work/openafs-1.4.8/src -O2 -pipe -fPIC -c curseswindows.c curseswindows.c:35:19: error: sgtty.h: No such file or directory *** Error code 1 ----- WBR & HTH -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From bsam at ipt.ru Fri Dec 12 06:23:01 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Fri Dec 12 06:23:07 2008 Subject: FreeBSD port for AFS client almost ready, a question though Message-ID: <49897500@bb.ipt.ru> Hello All, I managed to create a port which builds a kernel module for 7-STABLE. But before it is published I'd like to ask some questions. Here is the first one. Am I right to assume that all a client needs are a kernel module, an afsd daemon and an rc script? WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From bsam at ipt.ru Fri Dec 12 06:52:36 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Fri Dec 12 06:52:42 2008 Subject: OpenAFS on FreeBSD Progress (Works) In-Reply-To: <22B6C509EF7C4AB0A2D8350C31BB8D5D@valentine> (Tony Jago's message of "Tue\, 9 Dec 2008 09\:55\:16 +1000") References: <493ACAC4.5020806@linuxbox.com> <12501719@bb.ipt.ru> <493D898C.1030609@linuxbox.com> <22B6C509EF7C4AB0A2D8350C31BB8D5D@valentine> Message-ID: <57098597@bb.ipt.ru> "Tony Jago" writes: >>> I'd like to create two separate ports: openafs-server and >>> openafs-client. But I don't know right now how to do it. > > I would also recommend a openafs metaport that combines the client and > the server. What's the purpose of this port? I understand when a big collection of ports (like xorg, gnome, etc.) are combined together. > Have a look at the mysql-client, mysql-server and mysql > ports to see an example of how to do this sort of thing. I didn't manage to find a metaport for mysql... WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From matt at linuxbox.com Fri Dec 12 06:54:03 2008 From: matt at linuxbox.com (Matt Benjamin) Date: Fri Dec 12 06:54:10 2008 Subject: FreeBSD port for AFS client almost ready, a question though In-Reply-To: <49897500@bb.ipt.ru> References: <49897500@bb.ipt.ru> Message-ID: <49427B03.8070802@linuxbox.com> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 Hi Boris: Are... CellServDB ThisCell (maybe cacheinfo) /afs mount point /usr/vice/cache (or whatever mount point you'd prefer to use for the vice cache, should it be a vice cache) (and perhaps) krb5.conf (not _always_ required, but...) accounted for? Matt Boris Samorodov wrote: > Hello All, > > > I managed to create a port which builds a kernel module for 7-STABLE. > But before it is published I'd like to ask some questions. Here is > the first one. > > Am I right to assume that all a client needs are a kernel module, an > afsd daemon and an rc script? > > > WBR - -- Matt Benjamin The Linux Box 206 South Fifth Ave. Suite 150 Ann Arbor, MI 48104 http://linuxbox.com tel. 734-761-4689 fax. 734-769-8938 cel. 734-216-5309 -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.7 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFJQnsDJiSUUSaRdSURCBiWAJ9A6anC77ftrx/7euyVRDutVrg3mQCdG26G JelvgH+XpCx41TUPHaNxY/A= =cxtk -----END PGP SIGNATURE----- From bsam at ipt.ru Fri Dec 12 07:15:23 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Fri Dec 12 07:15:29 2008 Subject: FreeBSD port for AFS client almost ready, a question though In-Reply-To: <49427B03.8070802@linuxbox.com> (Matt Benjamin's message of "Fri\, 12 Dec 2008 09\:53\:55 -0500") References: <49897500@bb.ipt.ru> <49427B03.8070802@linuxbox.com> Message-ID: <14217222@bb.ipt.ru> Matt Benjamin writes: > Hi Boris: > > Are... > > CellServDB > ThisCell OK. > (maybe cacheinfo) Should this file be installed? Or should it be dynamically created? > /afs mount point That may be a check while installing and/or while loading kernel module/daemon. > /usr/vice/cache (or whatever mount point you'd prefer to use for the > vice cache, should it be a vice cache) A check while loading a daemon? > (and perhaps) > > krb5.conf (not _always_ required, but...) Hm. Not sure how to deal with this. Should it be a note after install? > accounted for? Sure, thanks. > Boris Samorodov wrote: >> Hello All, >> >> >> I managed to create a port which builds a kernel module for 7-STABLE. >> But before it is published I'd like to ask some questions. Here is >> the first one. >> >> Am I right to assume that all a client needs are a kernel module, an >> afsd daemon and an rc script? WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From girgen at FreeBSD.org Fri Dec 12 07:50:49 2008 From: girgen at FreeBSD.org (Palle Girgensohn) Date: Fri Dec 12 07:50:55 2008 Subject: OpenAFS on FreeBSD Progress (Works) In-Reply-To: <57098597@bb.ipt.ru> References: <493ACAC4.5020806@linuxbox.com> <12501719@bb.ipt.ru> <493D898C.1030609@linuxbox.com> <22B6C509EF7C4AB0A2D8350C31BB8D5D@valentine> <57098597@bb.ipt.ru> Message-ID: <3791F33A0694091B8F21252B@girgMac.local> --On fredag, fr 12 dec 2008 17.52.26 +0300 Boris Samorodov wrote: > "Tony Jago" writes: > >>>> I'd like to create two separate ports: openafs-server and >>>> openafs-client. But I don't know right now how to do it. >> >> I would also recommend a openafs metaport that combines the client and >> the server. > > What's the purpose of this port? I understand when a big collection > of ports (like xorg, gnome, etc.) are combined together. I two ports share the same code base, but install different files, then a metaport or similar could be clever. See for example databases/postgresql83-(server|client). The postgresql83-server depends on the client. The client can installed without the server. Similar to OpenAFS, hence. Now, they have the same source tarball, so the -client port just sets a few parameters and then uses the postgresql83-server port as a master Makefile. >> Have a look at the mysql-client, mysql-server and mysql >> ports to see an example of how to do this sort of thing. > > I didn't manage to find a metaport for mysql... Works almost exactly the same as described for postgresql. :-) Regards, Palle > > > WBR > -- > Boris Samorodov (bsam) > Research Engineer, http://www.ipt.ru Telephone & Internet SP > FreeBSD committer, http://www.FreeBSD.org The Power To Serve > _______________________________________________ > freebsd-afs@freebsd.org mailing list > http://lists.freebsd.org/mailman/listinfo/freebsd-afs > To unsubscribe, send any mail to "freebsd-afs-unsubscribe@freebsd.org" From bsam at ipt.ru Fri Dec 12 08:03:12 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Fri Dec 12 08:03:19 2008 Subject: OpenAFS on FreeBSD Progress (Works) In-Reply-To: <3791F33A0694091B8F21252B@girgMac.local> (Palle Girgensohn's message of "Fri\, 12 Dec 2008 16\:22\:55 +0100") References: <493ACAC4.5020806@linuxbox.com> <12501719@bb.ipt.ru> <493D898C.1030609@linuxbox.com> <22B6C509EF7C4AB0A2D8350C31BB8D5D@valentine> <57098597@bb.ipt.ru> <3791F33A0694091B8F21252B@girgMac.local> Message-ID: <92774356@bb.ipt.ru> Palle Girgensohn writes: > --On fredag, fr 12 dec 2008 17.52.26 +0300 Boris Samorodov > wrote: > >> "Tony Jago" writes: >> >>>>> I'd like to create two separate ports: openafs-server and >>>>> openafs-client. But I don't know right now how to do it. >>> >>> I would also recommend a openafs metaport that combines the client and >>> the server. >> >> What's the purpose of this port? I understand when a big collection >> of ports (like xorg, gnome, etc.) are combined together. > > I two ports share the same code base, but install different files, > then a metaport or similar could be clever. How? I don't undrestand what for we need a *third* port. I agree and was going to create two ports: server and client. > See for example > databases/postgresql83-(server|client). There is no third metaport for server and client with combines them. [1] > The postgresql83-server > depends on the client. The client can installed without the > server. Similar to OpenAFS, hence. Now, they have the same source > tarball, so the -client port just sets a few parameters and then uses > the postgresql83-server port as a master Makefile. You have just desribed what I was going to do (actually, has already done): create an openafs-sever (master) and openafs-client (a meta-port). >>> Have a look at the mysql-client, mysql-server and mysql >>> ports to see an example of how to do this sort of thing. >> >> I didn't manage to find a metaport for mysql... > > Works almost exactly the same as described for postgresql. :-) [1] ;-) WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From girgen at pingpong.net Fri Dec 12 08:07:28 2008 From: girgen at pingpong.net (Palle Girgensohn) Date: Fri Dec 12 08:07:34 2008 Subject: OpenAFS on FreeBSD Progress (Works) Message-ID: <2086E93584713946C07FC9AF@girgMac.local> --On fredag, fr 12 dec 2008 17.52.26 +0300 Boris Samorodov wrote: > "Tony Jago" writes: > >>>> I'd like to create two separate ports: openafs-server and >>>> openafs-client. But I don't know right now how to do it. >> >> I would also recommend a openafs metaport that combines the client and >> the server. > > What's the purpose of this port? I understand when a big collection > of ports (like xorg, gnome, etc.) are combined together. I two ports share the same code base, but install different files, then a metaport or similar could be clever. See for example databases/postgresql83-(server|client). The postgresql83-server depends on the client. The client can installed without the server. Similar to OpenAFS, hence. Now, they have the same source tarball, so the -client port just sets a few parameters and then uses the postgresql83-server port as a master Makefile. >> Have a look at the mysql-client, mysql-server and mysql >> ports to see an example of how to do this sort of thing. > > I didn't manage to find a metaport for mysql... Works almost exactly the same as described for postgresql. :-) Regards, Palle > > > WBR > -- > Boris Samorodov (bsam) > Research Engineer, http://www.ipt.ru Telephone & Internet SP > FreeBSD committer, http://www.FreeBSD.org The Power To Serve > _______________________________________________ > freebsd-afs@freebsd.org mailing list > http://lists.freebsd.org/mailman/listinfo/freebsd-afs > To unsubscribe, send any mail to "freebsd-afs-unsubscribe@freebsd.org" From girgen at pingpong.net Fri Dec 12 09:45:31 2008 From: girgen at pingpong.net (Palle Girgensohn) Date: Fri Dec 12 09:45:38 2008 Subject: OpenAFS on FreeBSD Progress (Works) In-Reply-To: <92774356@bb.ipt.ru> References: <493ACAC4.5020806@linuxbox.com> <12501719@bb.ipt.ru> <493D898C.1030609@linuxbox.com> <22B6C509EF7C4AB0A2D8350C31BB8D5D@valentine> <57098597@bb.ipt.ru> <3791F33A0694091B8F21252B@girgMac.local> <92774356@bb.ipt.ru> Message-ID: <6A11DD89CC7F53A7AFA7EBB9@girgMac.local> --On fredag, fr 12 dec 2008 19.03.07 +0300 Boris Samorodov wrote: > You have just desribed what I was going to do (actually, has already > done): create an openafs-sever (master) and openafs-client (a > meta-port). Sounds perfect. ;-) From bsam at ipt.ru Fri Dec 12 10:28:24 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Fri Dec 12 10:28:32 2008 Subject: OpenAFS on FreeBSD Progress (Works) In-Reply-To: <57098597@bb.ipt.ru> (Boris Samorodov's message of "Fri\, 12 Dec 2008 17\:52\:26 +0300") References: <493ACAC4.5020806@linuxbox.com> <12501719@bb.ipt.ru> <493D898C.1030609@linuxbox.com> <22B6C509EF7C4AB0A2D8350C31BB8D5D@valentine> <57098597@bb.ipt.ru> Message-ID: <26695644@bb.ipt.ru> Boris Samorodov writes: > "Tony Jago" writes: > >>>> I'd like to create two separate ports: openafs-server and >>>> openafs-client. But I don't know right now how to do it. >> >> I would also recommend a openafs metaport that combines the client and >> the server. > > What's the purpose of this port? I understand when a big collection > of ports (like xorg, gnome, etc.) are combined together. Replying to myself. Seems I found out why this port may be needed: Both server and client have common files (i.e. should be marked as have CONFLICTS and can't be installed with each other). WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From girgen at pingpong.net Fri Dec 12 15:34:59 2008 From: girgen at pingpong.net (Palle Girgensohn) Date: Fri Dec 12 15:35:05 2008 Subject: OpenAFS on FreeBSD Progress (Works) In-Reply-To: <26695644@bb.ipt.ru> References: <493ACAC4.5020806@linuxbox.com> <12501719@bb.ipt.ru> <493D898C.1030609@linuxbox.com> <22B6C509EF7C4AB0A2D8350C31BB8D5D@valentine> <57098597@bb.ipt.ru> <26695644@bb.ipt.ru> Message-ID: --On fredag, fr 12 dec 2008 21.28.19 +0300 Boris Samorodov wrote: > Boris Samorodov writes: >> "Tony Jago" writes: >> >>>>> I'd like to create two separate ports: openafs-server and >>>>> openafs-client. But I don't know right now how to do it. >>> >>> I would also recommend a openafs metaport that combines the client and >>> the server. >> >> What's the purpose of this port? I understand when a big collection >> of ports (like xorg, gnome, etc.) are combined together. > > Replying to myself. Seems I found out why this port may be needed: > Both server and client have common files (i.e. should be marked > as have CONFLICTS and can't be installed with each other). You don't need a meta port, you a client and a server port, and let the client install all common files, and let the server depend on the client, which is reasonable anyway. /Palle > > > WBR > -- > Boris Samorodov (bsam) > Research Engineer, http://www.ipt.ru Telephone & Internet SP > FreeBSD committer, http://www.FreeBSD.org The Power To Serve > _______________________________________________ > freebsd-afs@freebsd.org mailing list > http://lists.freebsd.org/mailman/listinfo/freebsd-afs > To unsubscribe, send any mail to "freebsd-afs-unsubscribe@freebsd.org" From minimarmot at gmail.com Fri Dec 12 16:10:58 2008 From: minimarmot at gmail.com (Ben Kaduk) Date: Fri Dec 12 16:11:04 2008 Subject: OpenAFS on FreeBSD Progress (Works) In-Reply-To: References: <493ACAC4.5020806@linuxbox.com> <12501719@bb.ipt.ru> <493D898C.1030609@linuxbox.com> <22B6C509EF7C4AB0A2D8350C31BB8D5D@valentine> <57098597@bb.ipt.ru> <26695644@bb.ipt.ru> Message-ID: <47d0403c0812121540u43349dedr96a8af0c63b3fe8a@mail.gmail.com> On Fri, Dec 12, 2008 at 6:34 PM, Palle Girgensohn wrote: > > > --On fredag, fr 12 dec 2008 21.28.19 +0300 Boris Samorodov > wrote: > >> Boris Samorodov writes: >>> >>> "Tony Jago" writes: >>> >>>>>> I'd like to create two separate ports: openafs-server and >>>>>> openafs-client. But I don't know right now how to do it. >>>> >>>> I would also recommend a openafs metaport that combines the client and >>>> the server. >>> >>> What's the purpose of this port? I understand when a big collection >>> of ports (like xorg, gnome, etc.) are combined together. >> >> Replying to myself. Seems I found out why this port may be needed: >> Both server and client have common files (i.e. should be marked >> as have CONFLICTS and can't be installed with each other). > > You don't need a meta port, you a client and a server port, and let the > client install all common files, and let the server depend on the client, > which is reasonable anyway. We have a couple of servers here (MIT SIPB) that aren't running clients. As the guy who installed them reasoned, it's just another thing that might go wrong. I think that separate client, server, and meta packages is perfectly reasonable. -Ben Kaduk From shadow at gmail.com Fri Dec 12 16:51:44 2008 From: shadow at gmail.com (Derrick Brashear) Date: Fri Dec 12 16:51:51 2008 Subject: OpenAFS on FreeBSD Progress (Works) In-Reply-To: References: <493ACAC4.5020806@linuxbox.com> <12501719@bb.ipt.ru> <493D898C.1030609@linuxbox.com> <22B6C509EF7C4AB0A2D8350C31BB8D5D@valentine> <57098597@bb.ipt.ru> <26695644@bb.ipt.ru> Message-ID: On Fri, Dec 12, 2008 at 6:34 PM, Palle Girgensohn wrote: > > > --On fredag, fr 12 dec 2008 21.28.19 +0300 Boris Samorodov > wrote: > > Boris Samorodov writes: >> >>> "Tony Jago" writes: >>> >>> I'd like to create two separate ports: openafs-server and >>>>>> openafs-client. But I don't know right now how to do it. >>>>>> >>>>> >>>> I would also recommend a openafs metaport that combines the client and >>>> the server. >>>> >>> >>> What's the purpose of this port? I understand when a big collection >>> of ports (like xorg, gnome, etc.) are combined together. >>> >> >> Replying to myself. Seems I found out why this port may be needed: >> Both server and client have common files (i.e. should be marked >> as have CONFLICTS and can't be installed with each other). >> > > You don't need a meta port, you a client and a server port, and let the > client install all common files, and let the server depend on the client, > which is reasonable anyway. > Given that an OpenAFS server can be run clientless I don't agree. From rick-freebsd2008 at kiwi-computer.com Fri Dec 12 17:02:58 2008 From: rick-freebsd2008 at kiwi-computer.com (Rick C. Petty) Date: Fri Dec 12 17:03:04 2008 Subject: OpenAFS on FreeBSD Progress (Works) In-Reply-To: References: <493ACAC4.5020806@linuxbox.com> <12501719@bb.ipt.ru> <493D898C.1030609@linuxbox.com> <22B6C509EF7C4AB0A2D8350C31BB8D5D@valentine> <57098597@bb.ipt.ru> <26695644@bb.ipt.ru> Message-ID: <20081213004251.GA88954@keira.kiwi-computer.com> On Sat, Dec 13, 2008 at 12:34:49AM +0100, Palle Girgensohn wrote: > > >Replying to myself. Seems I found out why this port may be needed: > >Both server and client have common files (i.e. should be marked > >as have CONFLICTS and can't be installed with each other). > > You don't need a meta port, you a client and a server port, and let the > client install all common files, and let the server depend on the client, > which is reasonable anyway. I think you have that backwards. The client contains a kernel module. The server is all run in userland. I think in this case, have the client depend upon the server and have the server install all common files. I agree that there should not be a metaport. -- Rick C. Petty From shadow at gmail.com Fri Dec 12 17:06:46 2008 From: shadow at gmail.com (Derrick Brashear) Date: Fri Dec 12 17:06:52 2008 Subject: OpenAFS on FreeBSD Progress (Works) In-Reply-To: <20081213004251.GA88954@keira.kiwi-computer.com> References: <493ACAC4.5020806@linuxbox.com> <12501719@bb.ipt.ru> <493D898C.1030609@linuxbox.com> <22B6C509EF7C4AB0A2D8350C31BB8D5D@valentine> <57098597@bb.ipt.ru> <26695644@bb.ipt.ru> <20081213004251.GA88954@keira.kiwi-computer.com> Message-ID: On Fri, Dec 12, 2008 at 7:42 PM, Rick C. Petty < rick-freebsd2008@kiwi-computer.com> wrote: > On Sat, Dec 13, 2008 at 12:34:49AM +0100, Palle Girgensohn wrote: > > > > >Replying to myself. Seems I found out why this port may be needed: > > >Both server and client have common files (i.e. should be marked > > >as have CONFLICTS and can't be installed with each other). > > > > You don't need a meta port, you a client and a server port, and let the > > client install all common files, and let the server depend on the client, > > which is reasonable anyway. > > I think you have that backwards. The client contains a kernel module. The > server is all run in userland. I think in this case, have the client > depend > upon the server and have the server install all common files. I agree that > there should not be a metaport. > > The server similarly does not depend on the client. You don't need the kernel module to run the server. -- Derrick From drosih at rpi.edu Fri Dec 12 18:29:07 2008 From: drosih at rpi.edu (Garance A Drosihn) Date: Fri Dec 12 18:29:13 2008 Subject: OpenAFS on FreeBSD Progress (Works) In-Reply-To: <47d0403c0812121540u43349dedr96a8af0c63b3fe8a@mail.gmail.com> References: <493ACAC4.5020806@linuxbox.com> <12501719@bb.ipt.ru> <493D898C.1030609@linuxbox.com> <22B6C509EF7C4AB0A2D8350C31BB8D5D@valentine> <57098597@bb.ipt.ru> <26695644@bb.ipt.ru> <47d0403c0812121540u43349dedr96a8af0c63b3fe8a@mail.gmail.com> Message-ID: At 6:40 PM -0500 12/12/08, Ben Kaduk wrote: >On Fri, Dec 12, 2008 at 6:34 PM, Palle Girgensohn wrote: >> > > You don't need a meta port, you a client and a server port, and let the >> client install all common files, and let the server depend on the client, >> which is reasonable anyway. > >We have a couple of servers here (MIT SIPB) that aren't running clients. >As the guy who installed them reasoned, it's just another thing that >might go wrong. > >I think that separate client, server, and meta packages is perfectly >reasonable. I'm pretty sure our AFS servers do not have the AFS client installed, either. Or at least, they did not, I'm not sure how they are set up right this minute. All our AFS clients are not servers, and all our AFS servers are not clients. (for cell /afs/rpi.edu ) -- Garance Alistair Drosehn = gad@gilead.netel.rpi.edu Senior Systems Programmer or gad@freebsd.org Rensselaer Polytechnic Institute or drosih@rpi.edu From tony at convoitec.com Fri Dec 12 18:42:28 2008 From: tony at convoitec.com (Tony Jago) Date: Fri Dec 12 18:42:35 2008 Subject: OpenAFS port [was: OpenAFS on FreeBSD Progress (Works)] In-Reply-To: References: <493ACAC4.5020806@linuxbox.com> <12501719@bb.ipt.ru> <493D898C.1030609@linuxbox.com> <22B6C509EF7C4AB0A2D8350C31BB8D5D@valentine> <57098597@bb.ipt.ru> <26695644@bb.ipt.ru> <20081213004251.GA88954@keira.kiwi-computer.com> Message-ID: I think that we probably don't need more then one port. Yes, I know I was the one what originally proposed the meta port but I have changed my mind :) The reason we had a server and a client port originally was that the server was the only bit working and the kernel model was set not to compile. The client was was arla client. Now that both the openafs server and client are supported by the openafs team I can see no reason why it shouldn't be all in one port. The port should have separate rc variable to allow the administrator to only start the client or the server if they choose to. openafs_client_enable="YES" and openafs_server_enable="YES" for example. This gets around all the conflicting file problems. The kernel module need only be loaded if the client is required. This would seem to be a much easier and cleaner solution. My 2 cents, Tony From rick-freebsd2008 at kiwi-computer.com Fri Dec 12 18:46:00 2008 From: rick-freebsd2008 at kiwi-computer.com (Rick C. Petty) Date: Fri Dec 12 18:46:07 2008 Subject: OpenAFS on FreeBSD Progress (Works) In-Reply-To: References: <493ACAC4.5020806@linuxbox.com> <12501719@bb.ipt.ru> <493D898C.1030609@linuxbox.com> <22B6C509EF7C4AB0A2D8350C31BB8D5D@valentine> <57098597@bb.ipt.ru> <26695644@bb.ipt.ru> <20081213004251.GA88954@keira.kiwi-computer.com> Message-ID: <20081213024559.GA89850@keira.kiwi-computer.com> On Fri, Dec 12, 2008 at 08:06:45PM -0500, Derrick Brashear wrote: > On Fri, Dec 12, 2008 at 7:42 PM, Rick C. Petty < > rick-freebsd2008@kiwi-computer.com> wrote: > > > > You don't need a meta port, you a client and a server port, and let the > > > client install all common files, and let the server depend on the client, > > > which is reasonable anyway. > > > > I think you have that backwards. The client contains a kernel module. The > > server is all run in userland. I think in this case, have the client > > depend > > upon the server and have the server install all common files. I agree that > > there should not be a metaport. > > > The server similarly does not depend on the client. You don't need the > kernel module to run the server. That's precisely what I said.. the server does not depend upon the client. Having the client depend on the server isn't that bad, unless the code is easily separable. Having the server depend upon the client would be bad. -- Rick C. Petty From girgen at pingpong.net Sat Dec 13 02:26:18 2008 From: girgen at pingpong.net (Palle Girgensohn) Date: Sat Dec 13 02:26:25 2008 Subject: OpenAFS port [was: OpenAFS on FreeBSD Progress (Works)] In-Reply-To: References: <493ACAC4.5020806@linuxbox.com> <12501719@bb.ipt.ru> <493D898C.1030609@linuxbox.com> <22B6C509EF7C4AB0A2D8350C31BB8D5D@valentine> <57098597@bb.ipt.ru> <26695644@bb.ipt.ru> <20081213004251.GA88954@keira.kiwi-computer.com> Message-ID: <42451957-717C-4CA3-97D9-E2ACABE55E34@pingpong.net> As long as nothing conflicts with arla, I also suggest an all in one installation. Keeps it simple, which is always important. Palle 13 dec 2008 kl. 03.27 skrev "Tony Jago" : > I think that we probably don't need more then one port. Yes, I know > I was the one what originally proposed the meta port but I have > changed my mind :) The reason we had a server and a client port > originally was that the server was the only bit working and the > kernel model was set not to compile. The client was was arla client. > Now that both the openafs server and client are supported by the > openafs team I can see no reason why it shouldn't be all in one > port. The port should have separate rc variable to allow the > administrator to only start the client or the server if they choose > to. openafs_client_enable="YES" and openafs_server_enable="YES" for > example. This gets around all the conflicting file problems. The > kernel module need only be loaded if the client is required. This > would seem to be a much easier and cleaner solution. > > My 2 cents, > > Tony From bsam at ipt.ru Sat Dec 13 05:19:47 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Sat Dec 13 05:19:55 2008 Subject: OpenAFS port In-Reply-To: <42451957-717C-4CA3-97D9-E2ACABE55E34@pingpong.net> (Palle Girgensohn's message of "Sat\, 13 Dec 2008 11\:26\:03 +0100") References: <493ACAC4.5020806@linuxbox.com> <12501719@bb.ipt.ru> <493D898C.1030609@linuxbox.com> <22B6C509EF7C4AB0A2D8350C31BB8D5D@valentine> <57098597@bb.ipt.ru> <26695644@bb.ipt.ru> <20081213004251.GA88954@keira.kiwi-computer.com> <42451957-717C-4CA3-97D9-E2ACABE55E34@pingpong.net> Message-ID: <60600083@bb.ipt.ru> Palle Girgensohn writes: > 13 dec 2008 kl. 03.27 skrev "Tony Jago" : > >> I think that we probably don't need more then one port. Yes, I know >> I was the one what originally proposed the meta port but I have >> changed my mind :) The reason we had a server and a client port >> originally was that the server was the only bit working and the >> kernel model was set not to compile. The client was was arla client. >> Now that both the openafs server and client are supported by the >> openafs team I can see no reason why it shouldn't be all in one >> port. The port should have separate rc variable to allow the >> administrator to only start the client or the server if they choose >> to. openafs_client_enable="YES" and openafs_server_enable="YES" for >> example. This gets around all the conflicting file problems. The >> kernel module need only be loaded if the client is required. This >> would seem to be a much easier and cleaner solution. > > As long as nothing conflicts with arla, I also suggest an all in one > installation. Keeps it simple, which is always important. I'd vote for that myself if and only if we speak about a ports subsystem. But there are packages as well. And for those who prefer using packages I'd rather give an opportunity. Said that I propose following ports: . net/openafs (server+client) . net/openafs-server; . net/openafs-client. One of them will be a master port (I don't figure out which one, but that will be either openafs or openafs-server). All of them will conflict each other, i.e. only one of them can be installed at a machine. That way we may give all users their chance. Opinions? Thanks! WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From girgen at pingpong.net Sat Dec 13 06:15:26 2008 From: girgen at pingpong.net (Palle Girgensohn) Date: Sat Dec 13 06:15:33 2008 Subject: OpenAFS port In-Reply-To: <60600083@bb.ipt.ru> References: <493ACAC4.5020806@linuxbox.com> <12501719@bb.ipt.ru> <493D898C.1030609@linuxbox.com> <22B6C509EF7C4AB0A2D8350C31BB8D5D@valentine> <57098597@bb.ipt.ru> <26695644@bb.ipt.ru> <20081213004251.GA88954@keira.kiwi-computer.com> <42451957-717C-4CA3-97D9-E2ACABE55E34@pingpong.net> <60600083@bb.ipt.ru> Message-ID: --On l?rdag, l? 13 dec 2008 16.19.40 +0300 Boris Samorodov wrote: > Palle Girgensohn writes: >> 13 dec 2008 kl. 03.27 skrev "Tony Jago" : >> >>> I think that we probably don't need more then one port. Yes, I know >>> I was the one what originally proposed the meta port but I have >>> changed my mind :) The reason we had a server and a client port >>> originally was that the server was the only bit working and the >>> kernel model was set not to compile. The client was was arla client. >>> Now that both the openafs server and client are supported by the >>> openafs team I can see no reason why it shouldn't be all in one >>> port. The port should have separate rc variable to allow the >>> administrator to only start the client or the server if they choose >>> to. openafs_client_enable="YES" and openafs_server_enable="YES" for >>> example. This gets around all the conflicting file problems. The >>> kernel module need only be loaded if the client is required. This >>> would seem to be a much easier and cleaner solution. >> >> As long as nothing conflicts with arla, I also suggest an all in one >> installation. Keeps it simple, which is always important. > > I'd vote for that myself if and only if we speak about a ports > subsystem. But there are packages as well. And for those who prefer > using packages I'd rather give an opportunity. > > Said that I propose following ports: > . net/openafs (server+client) > . net/openafs-server; > . net/openafs-client. > > One of them will be a master port (I don't figure out which one, > but that will be either openafs or openafs-server). All of them > will conflict each other, i.e. only one of them can be installed > at a machine. > > That way we may give all users their chance. > > Opinions? Thanks! Personally I thinks that is overly complicated. Installing a few binaries that I don't use is not a big deal. Having three ports for a server+client system is confusing, especially since it is not very common practice. I'd vote for one single port, with rc switches to activate the different parts. /Palle From bsam at ipt.ru Sat Dec 13 07:17:30 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Sat Dec 13 07:17:37 2008 Subject: OpenAFS port In-Reply-To: (Palle Girgensohn's message of "Sat\, 13 Dec 2008 15\:15\:19 +0100") References: <493ACAC4.5020806@linuxbox.com> <12501719@bb.ipt.ru> <493D898C.1030609@linuxbox.com> <22B6C509EF7C4AB0A2D8350C31BB8D5D@valentine> <57098597@bb.ipt.ru> <26695644@bb.ipt.ru> <20081213004251.GA88954@keira.kiwi-computer.com> <42451957-717C-4CA3-97D9-E2ACABE55E34@pingpong.net> <60600083@bb.ipt.ru> Message-ID: <46367482@bb.ipt.ru> Palle Girgensohn writes: > --On l?rdag, l? 13 dec 2008 16.19.40 +0300 Boris Samorodov > wrote: >> Palle Girgensohn writes: >>> 13 dec 2008 kl. 03.27 skrev "Tony Jago" : >>> >>>> I think that we probably don't need more then one port. Yes, I know >>>> I was the one what originally proposed the meta port but I have >>>> changed my mind :) The reason we had a server and a client port The more I work on the port the more I understand that it's not so trivial. I've changed my mind too. ;-) >>>> originally was that the server was the only bit working and the >>>> kernel model was set not to compile. The client was was arla client. >>>> Now that both the openafs server and client are supported by the >>>> openafs team I can see no reason why it shouldn't be all in one >>>> port. The port should have separate rc variable to allow the >>>> administrator to only start the client or the server if they choose >>>> to. openafs_client_enable="YES" and openafs_server_enable="YES" for >>>> example. This gets around all the conflicting file problems. The >>>> kernel module need only be loaded if the client is required. This >>>> would seem to be a much easier and cleaner solution. >>> >>> As long as nothing conflicts with arla, I also suggest an all in one >>> installation. Keeps it simple, which is always important. >> >> I'd vote for that myself if and only if we speak about a ports >> subsystem. But there are packages as well. And for those who prefer >> using packages I'd rather give an opportunity. >> >> Said that I propose following ports: >> . net/openafs (server+client) >> . net/openafs-server; >> . net/openafs-client. >> >> One of them will be a master port (I don't figure out which one, >> but that will be either openafs or openafs-server). All of them >> will conflict each other, i.e. only one of them can be installed >> at a machine. >> >> That way we may give all users their chance. >> >> Opinions? Thanks! > > Personally I thinks that is overly complicated. Installing a few > binaries that I don't use is not a big deal. Having three ports for a > server+client system is confusing, especially since it is not very > common practice. I'd vote for one single port, with rc switches to > activate the different parts. Anyway we can't create packages since kernel sources and objs are needed. Then really it's worth having only one port. The default net/openafs installs both server and client. While OPENAFS_SERVER_ONLY and OPENAFS_CLIENT_ONLY do what they should. WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From shadow at gmail.com Sat Dec 13 07:24:40 2008 From: shadow at gmail.com (Derrick Brashear) Date: Sat Dec 13 07:24:47 2008 Subject: OpenAFS port In-Reply-To: <60600083@bb.ipt.ru> References: <493ACAC4.5020806@linuxbox.com> <22B6C509EF7C4AB0A2D8350C31BB8D5D@valentine> <57098597@bb.ipt.ru> <26695644@bb.ipt.ru> <20081213004251.GA88954@keira.kiwi-computer.com> <42451957-717C-4CA3-97D9-E2ACABE55E34@pingpong.net> <60600083@bb.ipt.ru> Message-ID: I don't believe there are conflicts between client objects, and server objects. An independent client and server port should be possible. But if not, why a metaport and not say openafs-base that both client and server depend on? On Sat, Dec 13, 2008 at 8:19 AM, Boris Samorodov wrote: > Palle Girgensohn writes: > > 13 dec 2008 kl. 03.27 skrev "Tony Jago" : > > > >> I think that we probably don't need more then one port. Yes, I know > >> I was the one what originally proposed the meta port but I have > >> changed my mind :) The reason we had a server and a client port > >> originally was that the server was the only bit working and the > >> kernel model was set not to compile. The client was was arla client. > >> Now that both the openafs server and client are supported by the > >> openafs team I can see no reason why it shouldn't be all in one > >> port. The port should have separate rc variable to allow the > >> administrator to only start the client or the server if they choose > >> to. openafs_client_enable="YES" and openafs_server_enable="YES" for > >> example. This gets around all the conflicting file problems. The > >> kernel module need only be loaded if the client is required. This > >> would seem to be a much easier and cleaner solution. > > > > As long as nothing conflicts with arla, I also suggest an all in one > > installation. Keeps it simple, which is always important. > > I'd vote for that myself if and only if we speak about a ports > subsystem. But there are packages as well. And for those who prefer > using packages I'd rather give an opportunity. > > Said that I propose following ports: > . net/openafs (server+client) > . net/openafs-server; > . net/openafs-client. > > One of them will be a master port (I don't figure out which one, > but that will be either openafs or openafs-server). All of them > will conflict each other, i.e. only one of them can be installed > at a machine. > > That way we may give all users their chance. > > Opinions? Thanks! > > > WBR > -- > Boris Samorodov (bsam) > Research Engineer, http://www.ipt.ru Telephone & Internet SP > FreeBSD committer, http://www.FreeBSD.org The Power To Serve > -- Derrick From bsam at ipt.ru Sat Dec 13 07:34:56 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Sat Dec 13 07:35:03 2008 Subject: OpenAFS port In-Reply-To: (Derrick Brashear's message of "Sat\, 13 Dec 2008 10\:24\:39 -0500") References: <493ACAC4.5020806@linuxbox.com> <22B6C509EF7C4AB0A2D8350C31BB8D5D@valentine> <57098597@bb.ipt.ru> <26695644@bb.ipt.ru> <20081213004251.GA88954@keira.kiwi-computer.com> <42451957-717C-4CA3-97D9-E2ACABE55E34@pingpong.net> <60600083@bb.ipt.ru> Message-ID: <80286435@bb.ipt.ru> "Derrick Brashear" writes: > I don't believe there are conflicts between client objects, and server > objects. An independent client and server port should be possible. They both should install configuration files. Hence, they conflict here. > But if not, why a metaport and not say openafs-base that both client and > server depend on? Well, different ports have sense only if they can be packaged. If not, then use one port with apropriate options. WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From shadow at gmail.com Sat Dec 13 07:37:04 2008 From: shadow at gmail.com (Derrick Brashear) Date: Sat Dec 13 07:37:11 2008 Subject: OpenAFS port In-Reply-To: <80286435@bb.ipt.ru> References: <493ACAC4.5020806@linuxbox.com> <26695644@bb.ipt.ru> <20081213004251.GA88954@keira.kiwi-computer.com> <42451957-717C-4CA3-97D9-E2ACABE55E34@pingpong.net> <60600083@bb.ipt.ru> <80286435@bb.ipt.ru> Message-ID: On Sat, Dec 13, 2008 at 10:34 AM, Boris Samorodov wrote: > "Derrick Brashear" writes: > > > I don't believe there are conflicts between client objects, and server > > objects. An independent client and server port should be possible. > > They both should install configuration files. Hence, they conflict > here. > Server and client use separate configuration, unless it's stuff *you added*, and then it's a choice you made, OpenAFS itself doesn't have that issue. In the old transarc-style universe, /usr/afs/etc versus /usr/vice/etc From bsam at ipt.ru Sat Dec 13 07:44:11 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Sat Dec 13 07:44:18 2008 Subject: OpenAFS port In-Reply-To: (Derrick Brashear's message of "Sat\, 13 Dec 2008 10\:37\:03 -0500") References: <493ACAC4.5020806@linuxbox.com> <26695644@bb.ipt.ru> <20081213004251.GA88954@keira.kiwi-computer.com> <42451957-717C-4CA3-97D9-E2ACABE55E34@pingpong.net> <60600083@bb.ipt.ru> <80286435@bb.ipt.ru> Message-ID: <14205895@bb.ipt.ru> "Derrick Brashear" writes: > On Sat, Dec 13, 2008 at 10:34 AM, Boris Samorodov wrote: >> "Derrick Brashear" writes: >> >> > I don't believe there are conflicts between client objects, and server >> > objects. An independent client and server port should be possible. >> >> They both should install configuration files. Hence, they conflict >> here. > > Server and client use separate configuration, unless it's stuff *you added*, > and then it's a choice you made, OpenAFS itself doesn't have that issue. In > the old transarc-style universe, /usr/afs/etc versus /usr/vice/etc FreeBSD uses /usr/local/etc for all third-party configuration files, Both for servers and clients. So at least CellServDB and ThisCell should be dealed/installed by both. WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From shadow at gmail.com Sat Dec 13 07:50:38 2008 From: shadow at gmail.com (Derrick Brashear) Date: Sat Dec 13 07:50:45 2008 Subject: OpenAFS port In-Reply-To: <14205895@bb.ipt.ru> References: <493ACAC4.5020806@linuxbox.com> <20081213004251.GA88954@keira.kiwi-computer.com> <42451957-717C-4CA3-97D9-E2ACABE55E34@pingpong.net> <60600083@bb.ipt.ru> <80286435@bb.ipt.ru> <14205895@bb.ipt.ru> Message-ID: On Sat, Dec 13, 2008 at 10:43 AM, Boris Samorodov wrote: > "Derrick Brashear" writes: > > On Sat, Dec 13, 2008 at 10:34 AM, Boris Samorodov wrote: > >> "Derrick Brashear" writes: > >> > >> > I don't believe there are conflicts between client objects, and server > >> > objects. An independent client and server port should be possible. > >> > >> They both should install configuration files. Hence, they conflict > >> here. > > > > Server and client use separate configuration, unless it's stuff *you > added*, > > and then it's a choice you made, OpenAFS itself doesn't have that issue. > In > > the old transarc-style universe, /usr/afs/etc versus /usr/vice/etc > > FreeBSD uses /usr/local/etc for all third-party configuration files, > Both for servers and clients. So at least CellServDB and ThisCell > should be dealed/installed by both. > That's bad. Actually, it's horrible. It means you force a server to be in the same AFS cell as a client, whereas it's valid and I have done a client in one cell as a server in another. From bsam at ipt.ru Sat Dec 13 08:00:16 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Sat Dec 13 08:00:23 2008 Subject: OpenAFS port In-Reply-To: (Derrick Brashear's message of "Sat\, 13 Dec 2008 10\:50\:37 -0500") References: <493ACAC4.5020806@linuxbox.com> <20081213004251.GA88954@keira.kiwi-computer.com> <42451957-717C-4CA3-97D9-E2ACABE55E34@pingpong.net> <60600083@bb.ipt.ru> <80286435@bb.ipt.ru> <14205895@bb.ipt.ru> Message-ID: <48124915@bb.ipt.ru> "Derrick Brashear" writes: > On Sat, Dec 13, 2008 at 10:43 AM, Boris Samorodov wrote: >> "Derrick Brashear" writes: >> > On Sat, Dec 13, 2008 at 10:34 AM, Boris Samorodov wrote: >> >> "Derrick Brashear" writes: >> >> >> >> > I don't believe there are conflicts between client objects, and server >> >> > objects. An independent client and server port should be possible. >> >> >> >> They both should install configuration files. Hence, they conflict >> >> here. >> > >> > Server and client use separate configuration, unless it's stuff *you >> added*, >> > and then it's a choice you made, OpenAFS itself doesn't have that issue. >> In >> > the old transarc-style universe, /usr/afs/etc versus /usr/vice/etc >> >> FreeBSD uses /usr/local/etc for all third-party configuration files, >> Both for servers and clients. So at least CellServDB and ThisCell >> should be dealed/installed by both. > > That's bad. Actually, it's horrible. It means you force a server to be in > the same AFS cell as a client, whereas it's valid and I have done a client > in one cell as a server in another. Is it a typical configuration? OK, let's find out what we can do here. Is it possible to configure a client/server software to use non-default values? Say, ETCBASE/openafs-client/* and ETCBASE/openafs-server/*? But next, what to do with krb.conf? Seems there is no possibility to use different kerberos configuration file for client and server. WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From shadow at gmail.com Sat Dec 13 08:02:35 2008 From: shadow at gmail.com (Derrick Brashear) Date: Sat Dec 13 08:02:41 2008 Subject: OpenAFS port In-Reply-To: <48124915@bb.ipt.ru> References: <493ACAC4.5020806@linuxbox.com> <42451957-717C-4CA3-97D9-E2ACABE55E34@pingpong.net> <60600083@bb.ipt.ru> <80286435@bb.ipt.ru> <14205895@bb.ipt.ru> <48124915@bb.ipt.ru> Message-ID: > > > That's bad. Actually, it's horrible. It means you force a server to be in > > the same AFS cell as a client, whereas it's valid and I have done a > client > > in one cell as a server in another. > > Is it a typical configuration? > Doesn't matter. Your proposed configuration precludes it from working even in an atypical configuration, which means "yopu're doing it wrong". > > OK, let's find out what we can do here. Is it possible to configure > a client/server software to use non-default values? Say, > ETCBASE/openafs-client/* and ETCBASE/openafs-server/*? > You could do that. And now we're back to "can be installed independently", which was the point in the first place, since it's compile time, and you are either doing that, or not. > > But next, what to do with krb.conf? Seems there is no possibility > to use different kerberos configuration file for client and server. > Only the server uses a krb.conf at all. From minimarmot at gmail.com Sat Dec 13 08:10:25 2008 From: minimarmot at gmail.com (Ben Kaduk) Date: Sat Dec 13 08:10:36 2008 Subject: OpenAFS port In-Reply-To: References: <493ACAC4.5020806@linuxbox.com> <42451957-717C-4CA3-97D9-E2ACABE55E34@pingpong.net> <60600083@bb.ipt.ru> <80286435@bb.ipt.ru> <14205895@bb.ipt.ru> <48124915@bb.ipt.ru> Message-ID: <47d0403c0812130810r1bd04945k77101439bc97260@mail.gmail.com> On Sat, Dec 13, 2008 at 11:02 AM, Derrick Brashear wrote: >> >> >> OK, let's find out what we can do here. Is it possible to configure >> a client/server software to use non-default values? Say, >> ETCBASE/openafs-client/* and ETCBASE/openafs-server/*? >> > > You could do that. And now we're back to "can be installed independently", > which was the point in the first place, since it's compile time, and you are > either doing that, or not. I think I would prefer ETCBASE/openafs/client and ETCBASE/openafs/server, but either one is fine. This separation does mean that they can be installed independently, but I don't see anything in it that prevents _either_ option, the "one port with configure options for what to install" or the "separate ports for the separate client and server". -Ben Kaduk From shadow at gmail.com Sat Dec 13 08:15:55 2008 From: shadow at gmail.com (Derrick Brashear) Date: Sat Dec 13 08:16:02 2008 Subject: OpenAFS port In-Reply-To: <47d0403c0812130810r1bd04945k77101439bc97260@mail.gmail.com> References: <493ACAC4.5020806@linuxbox.com> <60600083@bb.ipt.ru> <80286435@bb.ipt.ru> <14205895@bb.ipt.ru> <48124915@bb.ipt.ru> <47d0403c0812130810r1bd04945k77101439bc97260@mail.gmail.com> Message-ID: On Sat, Dec 13, 2008 at 11:10 AM, Ben Kaduk wrote: > On Sat, Dec 13, 2008 at 11:02 AM, Derrick Brashear > wrote: > >> > >> > >> OK, let's find out what we can do here. Is it possible to configure > >> a client/server software to use non-default values? Say, > >> ETCBASE/openafs-client/* and ETCBASE/openafs-server/*? > >> > > > > You could do that. And now we're back to "can be installed > independently", > > which was the point in the first place, since it's compile time, and you > are > > either doing that, or not. > > I think I would prefer ETCBASE/openafs/client and > ETCBASE/openafs/server, but either one is fine. > > This separation does mean that they can be installed > independently, but I don't see anything in it that > prevents _either_ option, the "one port with configure > options for what to install" or the "separate ports for > the separate client and server". > Sorry, I didn't mean you were locked into 2 ports, but rather that once you got this far you were not locked out of it. > > -Ben Kaduk > -- Derrick From bsam at ipt.ru Sat Dec 13 08:23:32 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Sat Dec 13 08:23:39 2008 Subject: OpenAFS port In-Reply-To: (Derrick Brashear's message of "Sat\, 13 Dec 2008 11\:02\:33 -0500") References: <493ACAC4.5020806@linuxbox.com> <42451957-717C-4CA3-97D9-E2ACABE55E34@pingpong.net> <60600083@bb.ipt.ru> <80286435@bb.ipt.ru> <14205895@bb.ipt.ru> <48124915@bb.ipt.ru> Message-ID: <82043519@bb.ipt.ru> "Derrick Brashear" writes: >> > That's bad. Actually, it's horrible. It means you force a server to be in >> > the same AFS cell as a client, whereas it's valid and I have done a >> client >> > in one cell as a server in another. >> >> Is it a typical configuration? > > Doesn't matter. Your proposed configuration precludes it from working even > in an atypical configuration, which means "yopu're doing it wrong". That was a curious-like question, not defending my point of view. >> OK, let's find out what we can do here. Is it possible to configure >> a client/server software to use non-default values? Say, >> ETCBASE/openafs-client/* and ETCBASE/openafs-server/*? > > You could do that. And now we're back to "can be installed independently", > which was the point in the first place, since it's compile time, and you are > either doing that, or not. I'll do whatever the right thing is. ;-) That's why I ask. Seems that client may be configured while executing, i.e. changing only an rc script. >> But next, what to do with krb.conf? Seems there is no possibility >> to use different kerberos configuration file for client and server. > > Only the server uses a krb.conf at all. OK, good to know. Thanks for your help. WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From rick-freebsd2008 at kiwi-computer.com Sat Dec 13 09:20:21 2008 From: rick-freebsd2008 at kiwi-computer.com (Rick C. Petty) Date: Sat Dec 13 09:20:50 2008 Subject: OpenAFS port In-Reply-To: <46367482@bb.ipt.ru> References: <57098597@bb.ipt.ru> <26695644@bb.ipt.ru> <20081213004251.GA88954@keira.kiwi-computer.com> <42451957-717C-4CA3-97D9-E2ACABE55E34@pingpong.net> <60600083@bb.ipt.ru> <46367482@bb.ipt.ru> Message-ID: <20081213172019.GA96721@keira.kiwi-computer.com> On Sat, Dec 13, 2008 at 06:17:25PM +0300, Boris Samorodov wrote: > > > Personally I thinks that is overly complicated. Installing a few > > binaries that I don't use is not a big deal. Having three ports for a > > server+client system is confusing, especially since it is not very > > common practice. I'd vote for one single port, with rc switches to > > activate the different parts. > > Anyway we can't create packages since kernel sources and objs are > needed. Then really it's worth having only one port. The default > net/openafs installs both server and client. While OPENAFS_SERVER_ONLY > and OPENAFS_CLIENT_ONLY do what they should. I like the single-port idea the best, and use the OPENAFS_*_ONLY variables to disable one half. I agree with others that three ports is unnecessarily complicated. Unless there's really a technical reason to separate ports that can't be worked around, I vote: one port. Thank you for your efforts! -- Rick C. Petty From bsam at ipt.ru Sat Dec 13 09:38:01 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Sat Dec 13 09:38:13 2008 Subject: Configuration files structure for openafs port(s) Message-ID: <94529048@bb.ipt.ru> Hello List, here is a proposed coniguration files structure for openafs port(s): ----- PREFIX=/usr/local ETCBASE=${PREFIX}/etc AFSCONFDIR=${ETCBASE}/openafs AFSCLIENTCONFDIR=${AFSCONFDIR}/client #should this be named VICEETC? AFSSERVERCONFDIR=${AFSCONFDIR}/server #is there a historical variable? ----- WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From minimarmot at gmail.com Sat Dec 13 11:22:39 2008 From: minimarmot at gmail.com (Ben Kaduk) Date: Sat Dec 13 11:22:46 2008 Subject: Configuration files structure for openafs port(s) In-Reply-To: <94529048@bb.ipt.ru> References: <94529048@bb.ipt.ru> Message-ID: <47d0403c0812131122x82adfddu537d2be603869ad7@mail.gmail.com> On Sat, Dec 13, 2008 at 12:37 PM, Boris Samorodov wrote: > Hello List, > > > here is a proposed coniguration files structure for openafs port(s): > ----- > PREFIX=/usr/local > ETCBASE=${PREFIX}/etc > AFSCONFDIR=${ETCBASE}/openafs > AFSCLIENTCONFDIR=${AFSCONFDIR}/client #should this be named VICEETC? > AFSSERVERCONFDIR=${AFSCONFDIR}/server #is there a historical variable? > ----- That seems reasonable to me. I pulled your port from ftp://ftp.ipt.ru/pub/tmp/openafs.shar and I removed "--disable-kernel-module" from the Makefile on a recent -current system. The build will take a couple hours, but I did notice that we will probably need to add a --with-bsd-kernel-build line in the Makefile that defaults to /usr/obj/usr/src/sys/GENERIC but can be overridden for those of use with custom kernel configs. -Ben Kaduk From bsam at ipt.ru Sat Dec 13 11:44:08 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Sat Dec 13 11:44:16 2008 Subject: Configuration files structure for openafs port(s) In-Reply-To: <47d0403c0812131122x82adfddu537d2be603869ad7@mail.gmail.com> (Ben Kaduk's message of "Sat\, 13 Dec 2008 14\:22\:38 -0500") References: <94529048@bb.ipt.ru> <47d0403c0812131122x82adfddu537d2be603869ad7@mail.gmail.com> Message-ID: <94513977@bs1.sp34.ru> "Ben Kaduk" writes: > On Sat, Dec 13, 2008 at 12:37 PM, Boris Samorodov wrote: >> Hello List, >> >> >> here is a proposed coniguration files structure for openafs port(s): >> ----- >> PREFIX=/usr/local >> ETCBASE=${PREFIX}/etc >> AFSCONFDIR=${ETCBASE}/openafs >> AFSCLIENTCONFDIR=${AFSCONFDIR}/client #should this be named VICEETC? >> AFSSERVERCONFDIR=${AFSCONFDIR}/server #is there a historical variable? >> ----- > > That seems reasonable to me. OK. > I pulled your port from ftp://ftp.ipt.ru/pub/tmp/openafs.shar > and I removed "--disable-kernel-module" from the Makefile > on a recent -current system. I get an error on -current system when building client: ----- ... touch MODLOAD/sec_net.h cd MODLOAD ; echo make DESTDIR= single_compdir_libafs; make DESTDIR= single_compdir_libafs make DESTDIR= single_compdir_libafs ... cc -I. -I.. -I../nfs -I/work/a/ports/net/openafs/work/openafs-1.4.8/src -I/work/a/ports/net/openafs/work/openafs-1.4.8/src/afs -I/work/a/ports/net/openafs/work/openafs-1.4.8/src/afs/FBSD -I/work/a/ports/net/openafs/work/openafs-1.4.8/src/config -I/work/a/ports/net/openafs/work/openafs-1.4.8/src/rx/FBSD -I/work/a/ports/net/openafs/work/openafs-1.4.8/src/rxkad -I/work/a/ports/net/openafs/work/openafs-1.4.8/src/rxkad/domestic -I/work/a/ports/net/openafs/work/openafs-1.4.8/src/util -I/work/a/ports/net/openafs/work/openafs-1.4.8/src -I/work/a/ports/net/openafs/work/openafs-1.4.8/src/afs -I/work/a/ports/net/openafs/work/openafs-1.4.8/src/afs/FBSD -I/work/a/ports/net/openafs/work/openafs-1.4.8/src/util -I/work/a/ports/net/openafs/work/openafs-1.4.8/src/rxkad -I/work/a/ports/net/openafs/work/openafs-1.4.8/src/config -I/work/a/ports/net/openafs/work/openafs-1.4.8/src/fsint -I/work/a/ports/net/openafs/work/openafs-1.4.8/src/vlserver -I/work/a/ports/net/openafs/work/openafs-1.4.8/include -I/work/a/ports/net/openafs/work/openafs-1.4.8/include/afs -O -I. -I.. -I/work/a/ports/net/openafs/work/openafs-1.4.8/src/config -DAFSDEBUG -DKERNEL -DAFS -DVICE -DNFS -DUFS -DINET -DQUOTA -DGETMOUNT -Wall -nostdinc -I/usr/include -D_KERNEL -DKLD_MODULE -elf -mno-align-long-strings -fformat-extensions -fno-common -ffreestanding -I/a/ports/net/openafs/files -include opt_global.h -fno-strict-aliasing -O2 -o afs_vcache.o -c /work/a/ports/net/openafs/work/openafs-1.4.8/src/afs/afs_vcache.c /work/a/ports/net/openafs/work/openafs-1.4.8/src/afs/afs_vcache.c: In function 'afs_FlushReclaimedVcaches': /work/a/ports/net/openafs/work/openafs-1.4.8/src/afs/afs_vcache.c:602: warning: format '%x' expects type 'unsigned int', but argument 2 has type 'struct vcache *' /work/a/ports/net/openafs/work/openafs-1.4.8/src/afs/afs_vcache.c: In function 'afs_GetVCache': /work/a/ports/net/openafs/work/openafs-1.4.8/src/afs/afs_vcache.c:1840: warning: passing argument 3 of 'vinvalbuf' makes integer from pointer without a cast /work/a/ports/net/openafs/work/openafs-1.4.8/src/afs/afs_vcache.c:1840: error: too many arguments to function 'vinvalbuf' /work/a/ports/net/openafs/work/openafs-1.4.8/src/afs/afs_vcache.c: In function 'afs_NFSFindVCache': /work/a/ports/net/openafs/work/openafs-1.4.8/src/afs/afs_vcache.c:2820: warning: unused variable 'lock' *** Error code 1 Stop in /work/a/ports/net/openafs/work/openafs-1.4.8/src/libafs/MODLOAD. *** Error code 1 ----- > The build will take a couple hours, You should use and ancient machine... I's say that some 10-15 minutes are more reasonable. > but I did notice that we will probably need to add a > --with-bsd-kernel-build line in the Makefile that defaults to > /usr/obj/usr/src/sys/GENERIC but can be overridden for > those of use with custom kernel configs. Yes, I do it for experimental (not published yet) openafs port with client support. Thanks for your feedback. WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From minimarmot at gmail.com Sat Dec 13 12:00:46 2008 From: minimarmot at gmail.com (Ben Kaduk) Date: Sat Dec 13 12:00:55 2008 Subject: Configuration files structure for openafs port(s) In-Reply-To: <94513977@bs1.sp34.ru> References: <94529048@bb.ipt.ru> <47d0403c0812131122x82adfddu537d2be603869ad7@mail.gmail.com> <94513977@bs1.sp34.ru> Message-ID: <47d0403c0812131200w4fb9b4aekfb392e2356807826@mail.gmail.com> On Sat, Dec 13, 2008 at 2:44 PM, Boris Samorodov wrote: > "Ben Kaduk" writes: >> On Sat, Dec 13, 2008 at 12:37 PM, Boris Samorodov wrote: >>> Hello List, >>> >>> >>> here is a proposed coniguration files structure for openafs port(s): >>> ----- >>> PREFIX=/usr/local >>> ETCBASE=${PREFIX}/etc >>> AFSCONFDIR=${ETCBASE}/openafs >>> AFSCLIENTCONFDIR=${AFSCONFDIR}/client #should this be named VICEETC? >>> AFSSERVERCONFDIR=${AFSCONFDIR}/server #is there a historical variable? >>> ----- >> >> That seems reasonable to me. > > OK. > >> I pulled your port from ftp://ftp.ipt.ru/pub/tmp/openafs.shar >> and I removed "--disable-kernel-module" from the Makefile >> on a recent -current system. > > I get an error on -current system when building client: > ----- > ... > touch MODLOAD/sec_net.h > cd MODLOAD ; echo make DESTDIR= single_compdir_libafs; make DESTDIR= single_compdir_libafs > make DESTDIR= single_compdir_libafs > ... > cc -I. -I.. -I../nfs -I/work/a/ports/net/openafs/work/openafs-1.4.8/src -I/work/a/ports/net/openafs/work/openafs-1.4.8/src/afs -I/work/a/ports/net/openafs/work/openafs-1.4.8/src/afs/FBSD -I/work/a/ports/net/openafs/work/openafs-1.4.8/src/config -I/work/a/ports/net/openafs/work/openafs-1.4.8/src/rx/FBSD -I/work/a/ports/net/openafs/work/openafs-1.4.8/src/rxkad -I/work/a/ports/net/openafs/work/openafs-1.4.8/src/rxkad/domestic -I/work/a/ports/net/openafs/work/openafs-1.4.8/src/util -I/work/a/ports/net/openafs/work/openafs-1.4.8/src -I/work/a/ports/net/openafs/work/openafs-1.4.8/src/afs -I/work/a/ports/net/openafs/work/openafs-1.4.8/src/afs/FBSD -I/work/a/ports/net/openafs/work/openafs-1.4.8/src/util -I/work/a/ports/net/openafs/work/openafs-1.4.8/src/rxkad -I/work/a/ports/net/openafs/work/openafs-1.4.8/src/config -I/work/a/ports/net/openafs/work/openafs-1.4.8/src/fsint -I/work/a/ports/net/openafs/work/openafs-1.4.8/src/vlserver -I/work/a/ports/net/openafs/work/openafs-1.4.8/include -I/work/a/ports/net/openafs/work/openafs-1.4.8/include/afs -O -I. -I.. -I/work/a/ports/net/openafs/work/openafs-1.4.8/src/config -DAFSDEBUG -DKERNEL -DAFS -DVICE -DNFS -DUFS -DINET -DQUOTA -DGETMOUNT -Wall -nostdinc -I/usr/include -D_KERNEL -DKLD_MODULE -elf -mno-align-long-strings -fformat-extensions -fno-common -ffreestanding -I/a/ports/net/openafs/files -include opt_global.h -fno-strict-aliasing -O2 -o afs_vcache.o -c /work/a/ports/net/openafs/work/openafs-1.4.8/src/afs/afs_vcache.c > /work/a/ports/net/openafs/work/openafs-1.4.8/src/afs/afs_vcache.c: In function 'afs_FlushReclaimedVcaches': > /work/a/ports/net/openafs/work/openafs-1.4.8/src/afs/afs_vcache.c:602: warning: format '%x' expects type 'unsigned int', but argument 2 has type 'struct vcache *' > /work/a/ports/net/openafs/work/openafs-1.4.8/src/afs/afs_vcache.c: In function 'afs_GetVCache': > /work/a/ports/net/openafs/work/openafs-1.4.8/src/afs/afs_vcache.c:1840: warning: passing argument 3 of 'vinvalbuf' makes integer from pointer without a cast > /work/a/ports/net/openafs/work/openafs-1.4.8/src/afs/afs_vcache.c:1840: error: too many arguments to function 'vinvalbuf' > /work/a/ports/net/openafs/work/openafs-1.4.8/src/afs/afs_vcache.c: In function 'afs_NFSFindVCache': > /work/a/ports/net/openafs/work/openafs-1.4.8/src/afs/afs_vcache.c:2820: warning: unused variable 'lock' > *** Error code 1 > > Stop in /work/a/ports/net/openafs/work/openafs-1.4.8/src/libafs/MODLOAD. > *** Error code 1 > ----- > Yes, I see that the prototype of vinvalbuf has changed in current, recently --- the "curthread" argument should be removed. > >> but I did notice that we will probably need to add a >> --with-bsd-kernel-build line in the Makefile that defaults to >> /usr/obj/usr/src/sys/GENERIC but can be overridden for >> those of use with custom kernel configs. > > Yes, I do it for experimental (not published yet) openafs port > with client support. Ah, thanks! > > Thanks for your feedback. You're welcome; thanks for working on the port, I'm really excited about it. Hm, letting my compile continue, it seems that there will be more errors to deal with. I will try to batch future updates. -Ben Kaduk From bsam at ipt.ru Sat Dec 13 14:57:20 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Sat Dec 13 14:57:27 2008 Subject: Is afsd only a client file Message-ID: <37396849@bs1.sp34.ru> Hello List, seems that the file "afsd" belongs to client part. But it compiles and installs even when "--disable-kernel-module" option is used. Should I prevent this file to be installed by openafs-sever port? Maybe a better question: is there an option which prevents this file from building/installing? WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From minimarmot at gmail.com Sat Dec 13 18:24:38 2008 From: minimarmot at gmail.com (Ben Kaduk) Date: Sat Dec 13 18:24:44 2008 Subject: Changes to OpenAFS to support FreeBSD 8-CURRENT Message-ID: <47d0403c0812131824i489c50c6w456e62e43ecabdeb@mail.gmail.com> Hi Boris, To get a OpenAFS compiling on a couple-day-old current, I needed to change (in addition to the vinvalbuf change mentioned elsewhere) VSUID to S_ISUID and VSGID to S_ISGID in VNOPS/afs_vnop_attrs.c from this change: trasz 2008-09-10 13:16:41 UTC (note that S_ISUID is from stat.h and VUID was from vnode.h) remove the curthread argument from VOP_GETATTR() (two occurrences) in FBSD/osi_file.c:143 (this is under AFS_FBSD50_ENV ???) from this change: attilio 2008-08-28 15:23:18 UTC The vinvalbuf change also affects FBSD/osi_vm.c in two lines. change ap->a_mode to ap->a_accmode in FBSD/osi_vnodeops.c:691 and in the comment above. Arguably, this might be rather a bug in the vnode_if.h header where the structure is defined (libafs/MODLOAD/vnode_if.h, apparently). Now, everything compiles, but kldload is claiming that libafs.ko doesn't exist ... -Ben Kaduk From matt at linuxbox.com Sat Dec 13 18:56:15 2008 From: matt at linuxbox.com (Matt Benjamin) Date: Sat Dec 13 18:56:22 2008 Subject: Changes to OpenAFS to support FreeBSD 8-CURRENT In-Reply-To: <47d0403c0812131824i489c50c6w456e62e43ecabdeb@mail.gmail.com> References: <47d0403c0812131824i489c50c6w456e62e43ecabdeb@mail.gmail.com> Message-ID: <494475C2.7000303@linuxbox.com> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 Hi Ben, I got your previous email, but wasn't sure I could get to this. I haven't in fact been focused on 8-CURRENT, but on vnode locking and refcounting correctness across 7.0 to 8-CURRENT. (Making headway...) Anyway, I set that aside and whacked through a similar effort. The things you got sound right to me, though I'm surprised krx and afs_server.c compiled, but, it might have to do with whether you have network stack virtualization enabled. The module doesn't load presumably because of unresolved externals--here either the aforementioned (like in_ifaddrhead), and probably "suser(thread)" which is gone, replaced with "priv_check(thread)" (what makes me think that one won't be around long). Openafs-1.4.8 (openafs.org) with RT # 123933 compiles and loads, as above, but I have not run the code, it's just to block something in, ATM. Matt Ben Kaduk wrote: > Hi Boris, > > To get a OpenAFS compiling on a couple-day-old current, I needed to > change (in addition to > the vinvalbuf change mentioned elsewhere) > > VSUID to S_ISUID and VSGID to S_ISGID in VNOPS/afs_vnop_attrs.c > from this change: > trasz 2008-09-10 13:16:41 UTC > (note that S_ISUID is from stat.h and VUID was from vnode.h) > > remove the curthread argument from VOP_GETATTR() (two occurrences) > in FBSD/osi_file.c:143 (this is under AFS_FBSD50_ENV ???) > from this change: > attilio 2008-08-28 15:23:18 UTC > > The vinvalbuf change also affects FBSD/osi_vm.c in two lines. > > change ap->a_mode to ap->a_accmode in FBSD/osi_vnodeops.c:691 > and in the comment above. Arguably, this might be rather a bug > in the vnode_if.h header where the structure is defined > (libafs/MODLOAD/vnode_if.h, apparently). > > Now, everything compiles, but kldload is claiming that libafs.ko > doesn't exist ... > > -Ben Kaduk > _______________________________________________ > freebsd-afs@freebsd.org mailing list > http://lists.freebsd.org/mailman/listinfo/freebsd-afs > To unsubscribe, send any mail to "freebsd-afs-unsubscribe@freebsd.org" - -- Matt Benjamin The Linux Box 206 South Fifth Ave. Suite 150 Ann Arbor, MI 48104 http://linuxbox.com tel. 734-761-4689 fax. 734-769-8938 cel. 734-216-5309 -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.7 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFJRHXBJiSUUSaRdSURCIEPAJsFcRHTBEDsWs4y2r3h0LJlQ7LjzgCdGJS4 uFYwecbqcgAQeA1bXhSnqG4= =f20Q -----END PGP SIGNATURE----- From minimarmot at gmail.com Sat Dec 13 19:03:33 2008 From: minimarmot at gmail.com (Ben Kaduk) Date: Sat Dec 13 19:03:40 2008 Subject: Changes to OpenAFS to support FreeBSD 8-CURRENT In-Reply-To: <494475C2.7000303@linuxbox.com> References: <47d0403c0812131824i489c50c6w456e62e43ecabdeb@mail.gmail.com> <494475C2.7000303@linuxbox.com> Message-ID: <47d0403c0812131903m24969d40jb4300636edb18124@mail.gmail.com> On Sat, Dec 13, 2008 at 9:56 PM, Matt Benjamin wrote: > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA256 > > Hi Ben, > > I got your previous email, but wasn't sure I could get to this. I > haven't in fact been focused on 8-CURRENT, but on vnode locking and > refcounting correctness across 7.0 to 8-CURRENT. (Making headway...) > > Anyway, I set that aside and whacked through a similar effort. The > things you got sound right to me, though I'm surprised krx and > afs_server.c compiled, but, it might have to do with whether you have > network stack virtualization enabled. The module doesn't load > presumably because of unresolved externals--here either the > aforementioned (like in_ifaddrhead), and probably "suser(thread)" which > is gone, replaced with "priv_check(thread)" (what makes me think that > one won't be around long). Right. After a bit of googling, I thought to check the dmesg, where the helpful error link_elf: symbol suser undefined had appeared. Now I get to find where kldload is printing "No such file or directory" instead of something useful, and file a bug. > > Openafs-1.4.8 (openafs.org) with RT # 123933 compiles and loads, as > above, but I have not run the code, it's just to block something in, ATM. > Okay, I'll look at that some more. Thanks for taking the time out for me. -Ben Kaduk > Matt > > Ben Kaduk wrote: >> Hi Boris, >> >> To get a OpenAFS compiling on a couple-day-old current, I needed to >> change (in addition to >> the vinvalbuf change mentioned elsewhere) >> >> VSUID to S_ISUID and VSGID to S_ISGID in VNOPS/afs_vnop_attrs.c >> from this change: >> trasz 2008-09-10 13:16:41 UTC >> (note that S_ISUID is from stat.h and VUID was from vnode.h) >> >> remove the curthread argument from VOP_GETATTR() (two occurrences) >> in FBSD/osi_file.c:143 (this is under AFS_FBSD50_ENV ???) >> from this change: >> attilio 2008-08-28 15:23:18 UTC >> >> The vinvalbuf change also affects FBSD/osi_vm.c in two lines. >> >> change ap->a_mode to ap->a_accmode in FBSD/osi_vnodeops.c:691 >> and in the comment above. Arguably, this might be rather a bug >> in the vnode_if.h header where the structure is defined >> (libafs/MODLOAD/vnode_if.h, apparently). >> >> Now, everything compiles, but kldload is claiming that libafs.ko >> doesn't exist ... >> >> -Ben Kaduk >> _______________________________________________ >> freebsd-afs@freebsd.org mailing list >> http://lists.freebsd.org/mailman/listinfo/freebsd-afs >> To unsubscribe, send any mail to "freebsd-afs-unsubscribe@freebsd.org" > > > > > - -- > > Matt Benjamin > > The Linux Box > 206 South Fifth Ave. Suite 150 > Ann Arbor, MI 48104 > > http://linuxbox.com > > tel. 734-761-4689 > fax. 734-769-8938 > cel. 734-216-5309 > > -----BEGIN PGP SIGNATURE----- > Version: GnuPG v1.4.7 (GNU/Linux) > Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org > > iD8DBQFJRHXBJiSUUSaRdSURCIEPAJsFcRHTBEDsWs4y2r3h0LJlQ7LjzgCdGJS4 > uFYwecbqcgAQeA1bXhSnqG4= > =f20Q > -----END PGP SIGNATURE----- > From bsam at ipt.ru Sun Dec 14 06:22:30 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Sun Dec 14 06:22:47 2008 Subject: Is afsd only a client file In-Reply-To: (Tony Jago's message of "Sun\, 14 Dec 2008 09\:55\:01 +1000") References: <37396849@bs1.sp34.ru> Message-ID: <01705453@bs1.sp34.ru> Hello Tony, (restored CC to freebsd-afs@, maybe someone can push me to the right direction) "Tony Jago" writes: > The disable kernel module seems to only disable the kernel > module. There are a whole heap of commands that are only for > client. Infact the server bit is fairly small. Hm, until now I've been sure that client is kernel modile + afsd daemon + configuration files. Where can I find out which file should belong to which port? WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From matt at linuxbox.com Sun Dec 14 08:11:01 2008 From: matt at linuxbox.com (Matt Benjamin) Date: Sun Dec 14 08:11:08 2008 Subject: Is afsd only a client file In-Reply-To: <01705453@bs1.sp34.ru> References: <37396849@bs1.sp34.ru> <01705453@bs1.sp34.ru> Message-ID: <49453004.9010107@linuxbox.com> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 Hi, Boris Samorodov wrote: > "Tony Jago" writes: > >> The disable kernel module seems to only disable the kernel >> module. There are a whole heap of commands that are only for >> client. Infact the server bit is fairly small. > > Hm, until now I've been sure that client is kernel modile + > afsd daemon + configuration files. Where can I find out > which file should belong to which port? That is the client, minimally. Users need to get, list, and discard tokens (aklog, klog, unlog, tokens*), talk to the client to configure and debug it (fs), as well as do some AFS specific file operations (fs), manage groups (pts), possibly also users (pts). They should have vos, too (vos listvldb). You should also ship pagsh* (but I don't remember if pags work right now in the FreeBSD client). I'd think you should ship all of sys, rxgen, scout, udebug, cmdebug, rxdebug, fstrace, up, and bos for administrative and debugging purposes. Other distribution packages people have done (openafs.org) could be helpful, also openafs top-level make has a "dest" target that puts a Transarc style installation in a subtree "dest" of the openafs source tree being built. For the server, perhaps everything openafs would have put in its libexec or sbin, and not already mentioned (except maybe copyauth). And somehow you are accounting for configuration files and a var tree with appropiate permissions, as you noted. Matt - -- Matt Benjamin The Linux Box 206 South Fifth Ave. Suite 150 Ann Arbor, MI 48104 http://linuxbox.com tel. 734-761-4689 fax. 734-769-8938 cel. 734-216-5309 -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.7 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFJRTAEJiSUUSaRdSURCOn6AJ9a1eP1YnRIxvWLKOJ0CEAEUlJPlQCfXmIZ L1UB412oycjPWbILHY7kLxI= =qPHV -----END PGP SIGNATURE----- From matt at linuxbox.com Sun Dec 14 16:47:28 2008 From: matt at linuxbox.com (Matt Benjamin) Date: Sun Dec 14 16:47:42 2008 Subject: priv_check arg (oops) In-Reply-To: <494475C2.7000303@linuxbox.com> References: <47d0403c0812131824i489c50c6w456e62e43ecabdeb@mail.gmail.com> <494475C2.7000303@linuxbox.com> Message-ID: <4945A910.90700@linuxbox.com> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 RT # 123936 overlays, conditionally includes and passes the PRIV_ROOT argument as expected Matt Benjamin wrote: > > Openafs-1.4.8 (openafs.org) with RT # 123933 compiles and loads, as > above, but I have not run the code, it's just to block something in, ATM. - -- Matt Benjamin The Linux Box 206 South Fifth Ave. Suite 150 Ann Arbor, MI 48104 http://linuxbox.com tel. 734-761-4689 fax. 734-769-8938 cel. 734-216-5309 -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.7 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFJRahvJiSUUSaRdSURCF5+AJ9Ng0tzOYtFWsKiR08aDJkpCPiSTQCeIaaD ShQf6mK85MP/n4O/dhTuIOs= =bfj3 -----END PGP SIGNATURE----- From bsam at ipt.ru Mon Dec 15 04:52:20 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Mon Dec 15 04:52:30 2008 Subject: binary mtree for openafs port(s) In-Reply-To: <60571528@bs1.sp34.ru> (Boris Samorodov's message of "Mon\, 15 Dec 2008 14\:15\:19 +0300") References: <60571528@bs1.sp34.ru> Message-ID: <01693389@bb.ipt.ru> (resending after an hour timeout, seems my first letter went to nowhere) Hello List, this is time to decide which mtree for binaries is apropriate for openafs port(s). That's my proposal: PREFIX=/usr/local BINDIR=$PREFIX/bin SBINDIR=$PREFIX/sbin SBINDIR: all *server + upclient BINDIR: all other binaries What do you think? Thanks! WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From bsam at ipt.ru Mon Dec 15 06:36:21 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Mon Dec 15 06:36:28 2008 Subject: binary mtree for openafs port(s) Message-ID: <60571528@bs1.sp34.ru> Hello List, this is time to decide which mtree for binaries is apropriate for openafs port(s). That's my proposal: PREFIX=/usr/local BINDIR=$PREFIX/bin SBINDIR=$PREFIX/sbin SBINDIR: all *server + upclient BINDIR: all other binaries What do you think? Thanks! WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From elbarto at arcadebsd.org Mon Dec 15 15:24:44 2008 From: elbarto at arcadebsd.org (ElBarto) Date: Mon Dec 15 15:24:51 2008 Subject: binary mtree for openafs port(s) In-Reply-To: <01693389@bb.ipt.ru> References: <60571528@bs1.sp34.ru> <01693389@bb.ipt.ru> Message-ID: <4946E27E.1050803@arcadebsd.org> Boris Samorodov wrote: > (resending after an hour timeout, seems my first letter went to nowhere) > > Hello List, > > > this is time to decide which mtree for binaries is apropriate > for openafs port(s). That's my proposal: > > PREFIX=/usr/local > BINDIR=$PREFIX/bin > SBINDIR=$PREFIX/sbin > > SBINDIR: all *server + upclient > BINDIR: all other binaries > > What do you think? Thanks! > > > WBR > Hello, I'll prefer to have all the *server in $PREFIX/libexec/openafs like the pkgsrc for netbsd. As those binaries are not intend to be called manually it seems normal to not have them in /usr/local/bin or /usr/local/sbin. Just look at the pkgsrc, I kind of like the way it is done. -- Emmanuel Vadot elbarto@arcadebsd.org From bsam at ipt.ru Mon Dec 15 23:56:13 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Mon Dec 15 23:56:20 2008 Subject: binary mtree for openafs port(s) In-Reply-To: <4946E27E.1050803@arcadebsd.org> (elbarto@arcadebsd.org's message of "Tue\, 16 Dec 2008 00\:04\:30 +0100") References: <60571528@bs1.sp34.ru> <01693389@bb.ipt.ru> <4946E27E.1050803@arcadebsd.org> Message-ID: <49841540@bs1.sp34.ru> ElBarto writes: > Boris Samorodov wrote: >> this is time to decide which mtree for binaries is apropriate >> for openafs port(s). That's my proposal: >> >> PREFIX=/usr/local >> BINDIR=$PREFIX/bin >> SBINDIR=$PREFIX/sbin >> >> SBINDIR: all *server + upclient BINDIR: all other binaries >> >> What do you think? Thanks! > > I'll prefer to have all the *server in $PREFIX/libexec/openafs like > the pkgsrc for netbsd. That makes sence. > As those binaries are not intend to be called manually it seems normal > to not have them > in /usr/local/bin or /usr/local/sbin. > Just look at the pkgsrc, I kind of like the way it is done. This package contains the whole distribution. This is just what we do have now, but ee are going to separate server and client files. How can we do it? WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From bsam at ipt.ru Tue Dec 16 00:49:57 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Tue Dec 16 00:50:03 2008 Subject: openafs server executables (was: Re: Is afsd only a client file) In-Reply-To: <49453004.9010107@linuxbox.com> (Matt Benjamin's message of "Sun\, 14 Dec 2008 11\:10\:44 -0500") References: <37396849@bs1.sp34.ru> <01705453@bs1.sp34.ru> <49453004.9010107@linuxbox.com> Message-ID: <83768316_-_@bs1.sp34.ru> Matt Benjamin writes: >> Hm, until now I've been sure that client is kernel modile + >> afsd daemon + configuration files. Where can I find out >> which file should belong to which port? > > That is the client, minimally. Users need to get, list, and discard > tokens (aklog, klog, unlog, tokens*), talk to the client to configure > and debug it (fs), as well as do some AFS specific file operations (fs), > manage groups (pts), possibly also users (pts). They should have vos, > too (vos listvldb). You should also ship pagsh* (but I don't remember > if pags work right now in the FreeBSD client). > > I'd think you should ship all of sys, rxgen, scout, udebug, cmdebug, > rxdebug, fstrace, up, and bos for administrative and debugging purposes. > > Other distribution packages people have done (openafs.org) could be > helpful, also openafs top-level make has a "dest" target that puts a > Transarc style installation in a subtree "dest" of the openafs source > tree being built. > > For the server, perhaps everything openafs would have put in its libexec > or sbin, and not already mentioned (except maybe copyauth). And somehow > you are accounting for configuration files and a var tree with > appropiate permissions, as you noted. OK, seems that to determine the server part is an easier task. Let's say server executables are: ----- libexec/openafs/ buserver fileserver kaserver ptserver salvager upclient upserver vlserver volserver sbin/ bosserver ka-forwarder kadb_check kdb kpwvalid prdb_check pt_util rmtsysd vldb_check vldb_convert voldump volinfo vsys ----- And let's assume other executables belong to the client part. Any changes, additions, fixes? Thanks! WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From elbarto at arcadebsd.org Tue Dec 16 05:15:54 2008 From: elbarto at arcadebsd.org (ElBarto) Date: Tue Dec 16 05:16:01 2008 Subject: openafs server executables In-Reply-To: <83768316_-_@bs1.sp34.ru> References: <37396849@bs1.sp34.ru> <01705453@bs1.sp34.ru> <49453004.9010107@linuxbox.com> <83768316_-_@bs1.sp34.ru> Message-ID: <4947A9D0.6050900@arcadebsd.org> Boris Samorodov wrote: > Matt Benjamin writes: > > >>> Hm, until now I've been sure that client is kernel modile + >>> afsd daemon + configuration files. Where can I find out >>> which file should belong to which port? >>> >> That is the client, minimally. Users need to get, list, and discard >> tokens (aklog, klog, unlog, tokens*), talk to the client to configure >> and debug it (fs), as well as do some AFS specific file operations (fs), >> manage groups (pts), possibly also users (pts). They should have vos, >> too (vos listvldb). You should also ship pagsh* (but I don't remember >> if pags work right now in the FreeBSD client). >> >> I'd think you should ship all of sys, rxgen, scout, udebug, cmdebug, >> rxdebug, fstrace, up, and bos for administrative and debugging purposes. >> >> Other distribution packages people have done (openafs.org) could be >> helpful, also openafs top-level make has a "dest" target that puts a >> Transarc style installation in a subtree "dest" of the openafs source >> tree being built. >> >> For the server, perhaps everything openafs would have put in its libexec >> or sbin, and not already mentioned (except maybe copyauth). And somehow >> you are accounting for configuration files and a var tree with >> appropiate permissions, as you noted. >> > > OK, seems that to determine the server part is an easier task. > Let's say server executables are: > ----- > libexec/openafs/ > buserver > fileserver > kaserver > ptserver > salvager > upclient > upserver > vlserver > volserver > sbin/ > bosserver > ka-forwarder > kadb_check > kdb > kpwvalid > prdb_check > pt_util > rmtsysd > vldb_check > vldb_convert > voldump > volinfo > vsys > ----- > > And let's assume other executables belong to the client part. > > Any changes, additions, fixes? Thanks! > > > WBR > Well, the bos, vos and backup command are totally useless to an afs user so I think it will be better to included them in the openafs-server ports. I don't see either the "scout" or "udebug" command in your list, this is for monitoring servers so ... The uss command is only for the administrator. I think if your add those command in the list it should be good. -- Emmanuel Vadot elbarto@arcadebsd.org From bsam at ipt.ru Tue Dec 16 05:32:31 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Tue Dec 16 05:32:39 2008 Subject: openafs server executables In-Reply-To: <4947A9D0.6050900@arcadebsd.org> (elbarto@arcadebsd.org's message of "Tue\, 16 Dec 2008 14\:14\:56 +0100") References: <37396849@bs1.sp34.ru> <01705453@bs1.sp34.ru> <49453004.9010107@linuxbox.com> <83768316_-_@bs1.sp34.ru> <4947A9D0.6050900@arcadebsd.org> Message-ID: <69526901@bb.ipt.ru> ElBarto writes: > Boris Samorodov wrote: >> Matt Benjamin writes: >> >> >>>> Hm, until now I've been sure that client is kernel modile + >>>> afsd daemon + configuration files. Where can I find out >>>> which file should belong to which port? >>>> >>> That is the client, minimally. Users need to get, list, and discard >>> tokens (aklog, klog, unlog, tokens*), talk to the client to configure >>> and debug it (fs), as well as do some AFS specific file operations (fs), >>> manage groups (pts), possibly also users (pts). They should have vos, >>> too (vos listvldb). You should also ship pagsh* (but I don't remember >>> if pags work right now in the FreeBSD client). >>> >>> I'd think you should ship all of sys, rxgen, scout, udebug, cmdebug, >>> rxdebug, fstrace, up, and bos for administrative and debugging purposes. >>> >>> Other distribution packages people have done (openafs.org) could be >>> helpful, also openafs top-level make has a "dest" target that puts a >>> Transarc style installation in a subtree "dest" of the openafs source >>> tree being built. >>> >>> For the server, perhaps everything openafs would have put in its libexec >>> or sbin, and not already mentioned (except maybe copyauth). And somehow >>> you are accounting for configuration files and a var tree with >>> appropiate permissions, as you noted. >> >> OK, seems that to determine the server part is an easier task. >> Let's say server executables are: [old list was here] >> And let's assume other executables belong to the client part. >> >> Any changes, additions, fixes? Thanks! >> > Well, the bos, vos and backup command are totally useless to an afs > user so I think it will > be better to included them in the openafs-server ports. > I don't see either the "scout" or "udebug" command in your list, this > is for monitoring servers so ... > The uss command is only for the administrator. Done. > I think if your add those command in the list it should be good. Thanks for your help! Here is a changed list: ----- bin/ bos scout udebug libexec/openafs/ buserver fileserver kaserver ptserver salvager upclient upserver vlserver volserver sbin/ backup bosserver ka-forwarder kadb_check kdb kpwvalid prdb_check pt_util rmtsysd uss vldb_check vldb_convert voldump volinfo vos vsys ----- Any more fixes? WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From shadow at gmail.com Tue Dec 16 05:37:52 2008 From: shadow at gmail.com (Derrick Brashear) Date: Tue Dec 16 05:37:58 2008 Subject: openafs server executables (was: Re: Is afsd only a client file) In-Reply-To: <83768316_-_@bs1.sp34.ru> References: <37396849@bs1.sp34.ru> <01705453@bs1.sp34.ru> <49453004.9010107@linuxbox.com> <83768316_-_@bs1.sp34.ru> Message-ID: On Tue, Dec 16, 2008 at 3:49 AM, Boris Samorodov wrote: > Matt Benjamin writes: > > >> Hm, until now I've been sure that client is kernel modile + > >> afsd daemon + configuration files. Where can I find out > >> which file should belong to which port? > > > > That is the client, minimally. Users need to get, list, and discard > > tokens (aklog, klog, unlog, tokens*), talk to the client to configure > > and debug it (fs), as well as do some AFS specific file operations (fs), > > manage groups (pts), possibly also users (pts). They should have vos, > > too (vos listvldb). You should also ship pagsh* (but I don't remember > > if pags work right now in the FreeBSD client). > > > > I'd think you should ship all of sys, rxgen, scout, udebug, cmdebug, > > rxdebug, fstrace, up, and bos for administrative and debugging purposes. > > > > Other distribution packages people have done (openafs.org) could be > > helpful, also openafs top-level make has a "dest" target that puts a > > Transarc style installation in a subtree "dest" of the openafs source > > tree being built. > > > > For the server, perhaps everything openafs would have put in its libexec > > or sbin, and not already mentioned (except maybe copyauth). And somehow > > you are accounting for configuration files and a var tree with > > appropiate permissions, as you noted. > > OK, seems that to determine the server part is an easier task. > Let's say server executables are: > ----- > libexec/openafs/ > buserver > fileserver > kaserver > ptserver > salvager > upclient > upserver > vlserver > volserver > sbin/ > bosserver > ka-forwarder > kadb_check > kdb > kpwvalid > prdb_check > pt_util > rmtsysd > vldb_check > vldb_convert > voldump > volinfo > vsys > ----- > > And let's assume other executables belong to the client part. > Two copies of fs, vos, pts, bos, kas, asetkey, tokens, udebug, klog get installed normally. you want copies of at least fs, vos, bos, asetkey, udebug, pts (maybe not klog, tokens) in server, even if they are installed elsewhere in client. > > Any changes, additions, fixes? Thanks! > > > WBR > -- > Boris Samorodov (bsam) > Research Engineer, http://www.ipt.ru Telephone & Internet SP > FreeBSD committer, http://www.FreeBSD.org The Power To Serve > _______________________________________________ > freebsd-afs@freebsd.org mailing list > http://lists.freebsd.org/mailman/listinfo/freebsd-afs > To unsubscribe, send any mail to "freebsd-afs-unsubscribe@freebsd.org" > -- Derrick From elbarto at arcadebsd.org Tue Dec 16 05:41:54 2008 From: elbarto at arcadebsd.org (ElBarto) Date: Tue Dec 16 05:42:00 2008 Subject: binary mtree for openafs port(s) In-Reply-To: <49841540@bs1.sp34.ru> References: <60571528@bs1.sp34.ru> <01693389@bb.ipt.ru> <4946E27E.1050803@arcadebsd.org> <49841540@bs1.sp34.ru> Message-ID: <4947AFE8.3010506@arcadebsd.org> Boris Samorodov wrote: > ElBarto writes: > >> Boris Samorodov wrote: >> > > >>> this is time to decide which mtree for binaries is apropriate >>> for openafs port(s). That's my proposal: >>> >>> PREFIX=/usr/local >>> BINDIR=$PREFIX/bin >>> SBINDIR=$PREFIX/sbin >>> >>> SBINDIR: all *server + upclient BINDIR: all other binaries >>> >>> What do you think? Thanks! >>> >> I'll prefer to have all the *server in $PREFIX/libexec/openafs like >> the pkgsrc for netbsd. >> > > That makes sence. > > >> As those binaries are not intend to be called manually it seems normal >> to not have them >> in /usr/local/bin or /usr/local/sbin. >> Just look at the pkgsrc, I kind of like the way it is done. >> > > This package contains the whole distribution. This is just what we > do have now, but ee are going to separate server and client files. > How can we do it? > > > WBR > Well, For the client you only need all the binaries for mounting an afs server (don't know exactly which ones). This is the list I suggest : $PREFIX/sbin/afsd $PREFIX/bin/afsmonitor (seems more logic in the client part) $PREFIX/bin/dlog (don't really know what is it but seems a client part) $PREFIX/bin/dpass $PREFIX/bin/fs $PREFIX/sbin/fstrace $PREFIX/bin/klog $PREFIX/bin/knfs $PREFIX/bin/kpasswd $PREFIX/sbin/kpwvalid $PREFIX/bin/pagsh $PREFIX/bin/pts $PREFIX/sbin/rxdebug $PREFIX/bin/sys $PREFIX/bin/tokens $PREFIX/bin/translate_et $PREFIX/bin/unlog $PREFIX/bin/up $PREFIX/bin/xstat_cm_test The rest of the commands should be in the server ports. -- Emmanuel Vadot elbarto@arcadebsd.org From elbarto at arcadebsd.org Tue Dec 16 05:47:56 2008 From: elbarto at arcadebsd.org (ElBarto) Date: Tue Dec 16 05:48:02 2008 Subject: openafs server executables In-Reply-To: References: <37396849@bs1.sp34.ru> <01705453@bs1.sp34.ru> <49453004.9010107@linuxbox.com> <83768316_-_@bs1.sp34.ru> Message-ID: <4947B153.5060101@arcadebsd.org> Derrick Brashear wrote: > On Tue, Dec 16, 2008 at 3:49 AM, Boris Samorodov wrote: > > >> Matt Benjamin writes: >> >> >>>> Hm, until now I've been sure that client is kernel modile + >>>> afsd daemon + configuration files. Where can I find out >>>> which file should belong to which port? >>>> >>> That is the client, minimally. Users need to get, list, and discard >>> tokens (aklog, klog, unlog, tokens*), talk to the client to configure >>> and debug it (fs), as well as do some AFS specific file operations (fs), >>> manage groups (pts), possibly also users (pts). They should have vos, >>> too (vos listvldb). You should also ship pagsh* (but I don't remember >>> if pags work right now in the FreeBSD client). >>> >>> I'd think you should ship all of sys, rxgen, scout, udebug, cmdebug, >>> rxdebug, fstrace, up, and bos for administrative and debugging purposes. >>> >>> Other distribution packages people have done (openafs.org) could be >>> helpful, also openafs top-level make has a "dest" target that puts a >>> Transarc style installation in a subtree "dest" of the openafs source >>> tree being built. >>> >>> For the server, perhaps everything openafs would have put in its libexec >>> or sbin, and not already mentioned (except maybe copyauth). And somehow >>> you are accounting for configuration files and a var tree with >>> appropiate permissions, as you noted. >>> >> OK, seems that to determine the server part is an easier task. >> Let's say server executables are: >> ----- >> libexec/openafs/ >> buserver >> fileserver >> kaserver >> ptserver >> salvager >> upclient >> upserver >> vlserver >> volserver >> sbin/ >> bosserver >> ka-forwarder >> kadb_check >> kdb >> kpwvalid >> prdb_check >> pt_util >> rmtsysd >> vldb_check >> vldb_convert >> voldump >> volinfo >> vsys >> ----- >> >> And let's assume other executables belong to the client part. >> >> > > Two copies of fs, vos, pts, bos, kas, asetkey, tokens, udebug, klog get > installed normally. you want copies of at least fs, vos, bos, asetkey, > udebug, pts (maybe not klog, tokens) in server, even if they are installed > elsewhere in client. > So if I install openafs-client and openafs-server I'll have fs at two different place ? Seems ugly to me. As I said vos and bos are useless to an user so they should be in the server port. pts must be in the client port, even it's used for creating users, it can also be used to create sub-group (login:group). fs is needed to setting up a new cell but not after (server side), so it should be in the client ports. -- Emmanuel Vadot elbarto@arcadebsd.org From shadow at gmail.com Tue Dec 16 05:51:56 2008 From: shadow at gmail.com (Derrick Brashear) Date: Tue Dec 16 05:52:03 2008 Subject: openafs server executables In-Reply-To: <4947A9D0.6050900@arcadebsd.org> References: <37396849@bs1.sp34.ru> <01705453@bs1.sp34.ru> <49453004.9010107@linuxbox.com> <83768316_-_@bs1.sp34.ru> <4947A9D0.6050900@arcadebsd.org> Message-ID: > Well, the bos, vos and backup command are totally useless to an afs user so > I think it will as a mere mortal long before i had any admin power i used bos and vos. they are part of both the client and the server, and should be installed by both. > > be better to included them in the openafs-server ports. > I don't see either the "scout" or "udebug" command in your list, this is > for monitoring servers so ... > The uss command is only for the administrator. udebug is a debugging command, can be run by scripts etc. belongs in both. > > > I think if your add those command in the list it should be good. > > -- > Emmanuel Vadot > elbarto@arcadebsd.org > _______________________________________________ > freebsd-afs@freebsd.org mailing list > http://lists.freebsd.org/mailman/listinfo/freebsd-afs > To unsubscribe, send any mail to "freebsd-afs-unsubscribe@freebsd.org" > -- Derrick From shadow at gmail.com Tue Dec 16 05:54:58 2008 From: shadow at gmail.com (Derrick Brashear) Date: Tue Dec 16 05:55:05 2008 Subject: openafs server executables In-Reply-To: <4947B153.5060101@arcadebsd.org> References: <37396849@bs1.sp34.ru> <01705453@bs1.sp34.ru> <49453004.9010107@linuxbox.com> <83768316_-_@bs1.sp34.ru> <4947B153.5060101@arcadebsd.org> Message-ID: > > So if I install openafs-client and openafs-server I'll have fs at two > different place ? Seems ugly to me. Sounds subjective to me :) > > As I said vos and bos are useless to an user so they should be in the > server port. Repeating yourself doesn't make me more likely to agree. I suppose it depends how atuned to afs the user is but vos will check quotas where ACLs do not permit, (think help desk staffed by non admins) and bos can be used to verify that a server is unreachable because, say, it crashed, or what restart times are. (same idea) As a student I used them for those purposes uncompensated. -- Derrick From elbarto at arcadebsd.org Tue Dec 16 06:15:25 2008 From: elbarto at arcadebsd.org (ElBarto) Date: Tue Dec 16 06:15:30 2008 Subject: openafs server executables In-Reply-To: References: <37396849@bs1.sp34.ru> <01705453@bs1.sp34.ru> <49453004.9010107@linuxbox.com> <83768316_-_@bs1.sp34.ru> <4947B153.5060101@arcadebsd.org> Message-ID: <4947B7C4.8030004@arcadebsd.org> Derrick Brashear wrote: > > > So if I install openafs-client and openafs-server I'll have fs at > two different place ? Seems ugly to me. > > > Sounds subjective to me :) > What do you propose for the path of the redundant binaries ? > > As I said vos and bos are useless to an user so they should be in > the server port. > > > Repeating yourself doesn't make me more likely to agree. > > I suppose it depends how atuned to afs the user is but vos will check > quotas where ACLs do not permit, (think help desk staffed by non > admins) and bos can be used to verify that a server is unreachable > because, say, it crashed, or what restart times are. (same idea) > > As a student I used them for those purposes uncompensated. fs listquota is not enough for you ? (it's a real question, not ironic) I kind of disagree for splitting the openafs ports cause as you say some commands are useful (more or less) for both client and server part, but as it seems that it's going to be that way, I want the port to be clean and with no redundancy. The thing is you can't use the some vos command without a client (vos dump, restore etc ...) but those commands are not intend to users. Quoting the openafs administration reference : The commands in the vos command suite are the administrative interface to the Volume Server and Volume Location (VL) Server. System administrators use vos commands to create, move, delete, replicate, back up and examine volumes, among other operations. But I think you know that but know you see my point. -- Emmanuel Vadot elbarto@arcadebsd.org From matt at linuxbox.com Tue Dec 16 06:21:08 2008 From: matt at linuxbox.com (Matt Benjamin) Date: Tue Dec 16 06:21:14 2008 Subject: openafs server executables In-Reply-To: References: <37396849@bs1.sp34.ru> <01705453@bs1.sp34.ru> <49453004.9010107@linuxbox.com> <83768316_-_@bs1.sp34.ru> <4947B153.5060101@arcadebsd.org> Message-ID: <4947B946.2020004@linuxbox.com> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 Hi Guys, The "vos listvldb" is one of the first things I type when I logon to a strange AFS environment, and more importantly, one of the first things I ask someone to type if I have to debug a problem on a client. Both of the commands being discussed are the client program interfaces to server programs. Matt Derrick Brashear wrote: >> So if I install openafs-client and openafs-server I'll have fs at two >> different place ? Seems ugly to me. > > > Sounds subjective to me :) > > >> As I said vos and bos are useless to an user so they should be in the >> server port. > - -- Matt Benjamin The Linux Box 206 South Fifth Ave. Suite 150 Ann Arbor, MI 48104 http://linuxbox.com tel. 734-761-4689 fax. 734-769-8938 cel. 734-216-5309 -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.7 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFJR7lGJiSUUSaRdSURCLrbAJ0bvGCKJOIuWdrTFaCPNkTUBSDEWACdHiVm pbKq5e7EOTM9jW2rIb4oYFs= =pvGs -----END PGP SIGNATURE----- From shadow at gmail.com Tue Dec 16 06:28:22 2008 From: shadow at gmail.com (Derrick Brashear) Date: Tue Dec 16 06:28:29 2008 Subject: openafs server executables In-Reply-To: <4947B7C4.8030004@arcadebsd.org> References: <37396849@bs1.sp34.ru> <01705453@bs1.sp34.ru> <49453004.9010107@linuxbox.com> <83768316_-_@bs1.sp34.ru> <4947B153.5060101@arcadebsd.org> <4947B7C4.8030004@arcadebsd.org> Message-ID: On Tue, Dec 16, 2008 at 9:14 AM, ElBarto wrote: > Derrick Brashear wrote: > >> >> >> So if I install openafs-client and openafs-server I'll have fs at >> two different place ? Seems ugly to me. >> >> >> Sounds subjective to me :) >> >> > > What do you propose for the path of the redundant binaries ? > Putting the extra copies in sbin > > > >> As I said vos and bos are useless to an user so they should be in >> the server port. >> >> >> Repeating yourself doesn't make me more likely to agree. >> >> I suppose it depends how atuned to afs the user is but vos will check >> quotas where ACLs do not permit, (think help desk staffed by non admins) and >> bos can be used to verify that a server is unreachable because, say, it >> crashed, or what restart times are. (same idea) >> >> As a student I used them for those purposes uncompensated. >> > > fs listquota is not enough for you ? (it's a real question, not ironic) fs lq /afs/dementia.org/usr/shadow and tell me the quota. Then vos examine user.shadow -c dementia.org > I kind of disagree for splitting the openafs ports cause as you say some > commands are useful (more or less) for both client and server part, but as > it seems that it's going to be that way, I want the port to be clean and > with no redundancy. > The thing is you can't use the some vos command without a client (vos dump, > restore etc ...) but those commands are not intend to users. vos dump can be used without a client. > > Quoting the openafs administration reference : > > The commands in the vos command suite are the administrative interface to > the Volume Server and Volume Location (VL) Server. System administrators use > vos commands to create, move, delete, replicate, back up and examine > volumes, among other operations. > > But I think you know that but know you see my point. > You could say that I am intimately versed in the functionality of the pieces and their intended and actual purposes. > > > -- > Emmanuel Vadot > elbarto@arcadebsd.org > > -- Derrick From girgen at pingpong.net Tue Dec 16 06:46:34 2008 From: girgen at pingpong.net (Palle Girgensohn) Date: Tue Dec 16 06:46:42 2008 Subject: openafs server executables In-Reply-To: References: <37396849@bs1.sp34.ru> <01705453@bs1.sp34.ru> <49453004.9010107@linuxbox.com> <83768316_-_@bs1.sp34.ru> <4947B153.5060101@arcadebsd.org> <4947B7C4.8030004@arcadebsd.org> Message-ID: <1B1831E49644E24747EAC2A9@girgMac.local> --On tisdag, ti 16 dec 2008 09.28.21 -0500 Derrick Brashear wrote: > On Tue, Dec 16, 2008 at 9:14 AM, ElBarto wrote: > >> Derrick Brashear wrote: >> >>> >>> >>> So if I install openafs-client and openafs-server I'll have fs at >>> two different place ? Seems ugly to me. >>> >>> >>> Sounds subjective to me :) >>> >>> >> >> What do you propose for the path of the redundant binaries ? >> > > Putting the extra copies in sbin This is stupid. Why two copies? What if server and client installations get out of sync? That couldmake problems very hard to debug, since it is hardly really well specified which of bin and sbin that comes first in the path. The user will get confused, at best. It is really a very good argument for just having one port. I see no real reason why there should be a separation. Some client binaries are only needed when the server is installed, hence they need both. There seems to be no clean cut between server and client in the upstream sources (client-only and server-only installations share some binaries, e.g.), then we should not force such a (non-clean) cut into the FreeBSD ports system either. Palle From bsam at ipt.ru Tue Dec 16 06:56:46 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Tue Dec 16 06:56:52 2008 Subject: openafs port(s) in general (was: Re: openafs server executables) In-Reply-To: <4947B7C4.8030004@arcadebsd.org> (elbarto@arcadebsd.org's message of "Tue\, 16 Dec 2008 15\:14\:28 +0100") References: <37396849@bs1.sp34.ru> <01705453@bs1.sp34.ru> <49453004.9010107@linuxbox.com> <83768316_-_@bs1.sp34.ru> <4947B153.5060101@arcadebsd.org> <4947B7C4.8030004@arcadebsd.org> Message-ID: <35608451_-_@bb.ipt.ru> ElBarto writes: > I kind of disagree for splitting the openafs ports cause as you say > some commands are useful (more or less) for both client and server > part, but as it seems that it's going to be that way, I want the port > to be clean and with no redundancy. Let me elaborate on this a little. First off, how the port will look like is a matter of discussion. I'm open to suggestions. As well as someone may take the port I've done and suggest his version. ;-) My intention is following. The main port (say, net/openafs) will be one and will install the whole thing. And there will be some options (say WITH_SERVER_ONLY, WITH_CLIENT_ONLY) which will allow installing the relevant part. (BTW in this case who do care if a slave ports like net/openafs-server and/or openafs-client exist?) While discussing port(s) questions I was told that server and client parts are totally different and don't have anything in common. And if they do have then my ports are wrong. That and only that pushed me to find out which file belongs to which part. If you (read: developers, users and admins of OpenAFS on FreeBSD) give me a definite answer then I'll create such a port. You may even give a triple division: those files are server, those file are client, while those are common! I'll deal with it. Thanks. WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From bsam at ipt.ru Tue Dec 16 07:44:52 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Tue Dec 16 07:44:59 2008 Subject: openafs port(s) in general In-Reply-To: <35608451_-_@bb.ipt.ru> (Boris Samorodov's message of "Tue\, 16 Dec 2008 17\:56\:44 +0300") References: <37396849@bs1.sp34.ru> <01705453@bs1.sp34.ru> <49453004.9010107@linuxbox.com> <83768316_-_@bs1.sp34.ru> <4947B153.5060101@arcadebsd.org> <4947B7C4.8030004@arcadebsd.org> <35608451_-_@bb.ipt.ru> Message-ID: <69525564@bb.ipt.ru> Boris Samorodov writes: > If you (read: developers, users and admins of OpenAFS on FreeBSD) give > me a definite answer then I'll create such a port. Hm, I've found an answer. This page seems to be authorative enough: http://www.openafs.org/manpages/index.html WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From shadow at gmail.com Tue Dec 16 08:19:30 2008 From: shadow at gmail.com (Derrick Brashear) Date: Tue Dec 16 08:19:36 2008 Subject: openafs port(s) in general In-Reply-To: <69525564@bb.ipt.ru> References: <37396849@bs1.sp34.ru> <01705453@bs1.sp34.ru> <49453004.9010107@linuxbox.com> <83768316_-_@bs1.sp34.ru> <4947B153.5060101@arcadebsd.org> <4947B7C4.8030004@arcadebsd.org> <35608451_-_@bb.ipt.ru> <69525564@bb.ipt.ru> Message-ID: On Tue, Dec 16, 2008 at 10:44 AM, Boris Samorodov wrote: > Boris Samorodov writes: > > > If you (read: developers, users and admins of OpenAFS on FreeBSD) give > > me a definite answer then I'll create such a port. > > Hm, I've found an answer. This page seems to be authorative enough: > http://www.openafs.org/manpages/index.html > > notice some commands (vos) appear in "user" and "administrator"... From bsam at ipt.ru Tue Dec 16 08:26:21 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Tue Dec 16 08:26:28 2008 Subject: openafs port(s) in general In-Reply-To: (Derrick Brashear's message of "Tue\, 16 Dec 2008 11\:19\:29 -0500") References: <37396849@bs1.sp34.ru> <01705453@bs1.sp34.ru> <49453004.9010107@linuxbox.com> <83768316_-_@bs1.sp34.ru> <4947B153.5060101@arcadebsd.org> <4947B7C4.8030004@arcadebsd.org> <35608451_-_@bb.ipt.ru> <69525564@bb.ipt.ru> Message-ID: <03443077@bb.ipt.ru> "Derrick Brashear" writes: > On Tue, Dec 16, 2008 at 10:44 AM, Boris Samorodov wrote: >> Boris Samorodov writes: >> >> > If you (read: developers, users and admins of OpenAFS on FreeBSD) give >> > me a definite answer then I'll create such a port. >> >> Hm, I've found an answer. This page seems to be authorative enough: >> http://www.openafs.org/manpages/index.html >> > notice some commands (vos) appear in "user" and "administrator"... Didn't find. Which one? WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From shadow at gmail.com Tue Dec 16 08:30:30 2008 From: shadow at gmail.com (Derrick Brashear) Date: Tue Dec 16 08:30:37 2008 Subject: openafs port(s) in general In-Reply-To: <03443077@bb.ipt.ru> References: <37396849@bs1.sp34.ru> <83768316_-_@bs1.sp34.ru> <4947B153.5060101@arcadebsd.org> <4947B7C4.8030004@arcadebsd.org> <35608451_-_@bb.ipt.ru> <69525564@bb.ipt.ru> <03443077@bb.ipt.ru> Message-ID: On Tue, Dec 16, 2008 at 11:26 AM, Boris Samorodov wrote: > "Derrick Brashear" writes: > > On Tue, Dec 16, 2008 at 10:44 AM, Boris Samorodov wrote: > >> Boris Samorodov writes: > >> > >> > If you (read: developers, users and admins of OpenAFS on FreeBSD) give > >> > me a definite answer then I'll create such a port. > >> > >> Hm, I've found an answer. This page seems to be authorative enough: > >> http://www.openafs.org/manpages/index.html > >> > > notice some commands (vos) appear in "user" and "administrator"... > > Didn't find. Which one? > actually i guess vos is always "user" according to that, which is bunk. you must be an admin to run e.g. vos changeloc, vos addsite, etc... whereas vos examine, not so much. i could edit the web page in question if it'd help ;) but really what it means is vos_* appears in man1, and the "admin" stuff appears in man8. it's sort of arbitrary. From bsam at ipt.ru Tue Dec 16 08:49:04 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Tue Dec 16 08:49:10 2008 Subject: openafs port(s) in general In-Reply-To: (Derrick Brashear's message of "Tue\, 16 Dec 2008 11\:30\:29 -0500") References: <37396849@bs1.sp34.ru> <83768316_-_@bs1.sp34.ru> <4947B153.5060101@arcadebsd.org> <4947B7C4.8030004@arcadebsd.org> <35608451_-_@bb.ipt.ru> <69525564@bb.ipt.ru> <03443077@bb.ipt.ru> Message-ID: <37361713@bb.ipt.ru> "Derrick Brashear" writes: > i could edit the web page in question if it'd help ;) Yes, please. WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From minimarmot at gmail.com Tue Dec 16 09:10:39 2008 From: minimarmot at gmail.com (Ben Kaduk) Date: Tue Dec 16 09:10:48 2008 Subject: openafs server executables In-Reply-To: <1B1831E49644E24747EAC2A9@girgMac.local> References: <37396849@bs1.sp34.ru> <01705453@bs1.sp34.ru> <49453004.9010107@linuxbox.com> <83768316_-_@bs1.sp34.ru> <4947B153.5060101@arcadebsd.org> <4947B7C4.8030004@arcadebsd.org> <1B1831E49644E24747EAC2A9@girgMac.local> Message-ID: <47d0403c0812160910j92b5514y79652ee889e47423@mail.gmail.com> On Tue, Dec 16, 2008 at 9:46 AM, Palle Girgensohn wrote: > > > --On tisdag, ti 16 dec 2008 09.28.21 -0500 Derrick Brashear > wrote: > >> On Tue, Dec 16, 2008 at 9:14 AM, ElBarto wrote: >> >>> Derrick Brashear wrote: >>> >>>> >>>> >>>> So if I install openafs-client and openafs-server I'll have fs at >>>> two different place ? Seems ugly to me. >>>> >>>> >>>> Sounds subjective to me :) >>>> >>>> >>> >>> What do you propose for the path of the redundant binaries ? >>> >> >> Putting the extra copies in sbin > > This is stupid. Why two copies? What if server and client installations get > out of sync? That couldmake problems very hard to debug, since it is hardly > really well specified which of bin and sbin that comes first in the path. > The user will get confused, at best. > > It is really a very good argument for just having one port. I see no real > reason why there should be a separation. Some client binaries are only > needed when the server is installed, hence they need both. There seems to be > no clean cut between server and client in the upstream sources (client-only > and server-only installations share some binaries, e.g.), then we should not > force such a (non-clean) cut into the FreeBSD ports system either. There are some things that are clearly client, and some that are clearly server, and some that are shared. Now, we could probably argue for a while about exactly which ones are which, but that doesn't need to affect a conceptual breakdown into one or more FreeBSD ports. The three-way division mentioned elsewhere is, in fact, admirably suited to this situation: openafs-base for the shared commands openafs-server (DEPENDS openafs-base) for the server openafs-client (also DEPENDS openafs-base) is the client And one could also make a meta-package openafs that depends on both client and server, if one wants. The concern about having two different versions of fs, etc. is gone; one can even use the version-dependency tracking of FreeBSD ports to insure that the client and server utilities do not get out of sync. -Ben Kaduk From bsam at ipt.ru Tue Dec 16 09:38:47 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Tue Dec 16 09:38:54 2008 Subject: openafs server executables In-Reply-To: <47d0403c0812160910j92b5514y79652ee889e47423@mail.gmail.com> (Ben Kaduk's message of "Tue\, 16 Dec 2008 12\:10\:38 -0500") References: <37396849@bs1.sp34.ru> <01705453@bs1.sp34.ru> <49453004.9010107@linuxbox.com> <83768316_-_@bs1.sp34.ru> <4947B153.5060101@arcadebsd.org> <4947B7C4.8030004@arcadebsd.org> <1B1831E49644E24747EAC2A9@girgMac.local> <47d0403c0812160910j92b5514y79652ee889e47423@mail.gmail.com> Message-ID: <71288730@bb.ipt.ru> "Ben Kaduk" writes: > The three-way division mentioned elsewhere is, in fact, admirably > suited to this situation: > openafs-base for the shared commands > openafs-server (DEPENDS openafs-base) for the server > openafs-client (also DEPENDS openafs-base) is the client Yes, I meant this when I'd said I could deal with triplex division. Who can actually show the division? WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From shadow at gmail.com Tue Dec 16 09:56:43 2008 From: shadow at gmail.com (Derrick Brashear) Date: Tue Dec 16 09:56:50 2008 Subject: openafs port(s) in general In-Reply-To: <37361713@bb.ipt.ru> References: <37396849@bs1.sp34.ru> <4947B153.5060101@arcadebsd.org> <4947B7C4.8030004@arcadebsd.org> <35608451_-_@bb.ipt.ru> <69525564@bb.ipt.ru> <03443077@bb.ipt.ru> <37361713@bb.ipt.ru> Message-ID: On Tue, Dec 16, 2008 at 11:49 AM, Boris Samorodov wrote: > "Derrick Brashear" writes: > > > i could edit the web page in question if it'd help ;) > > Yes, please. > Actually, it's the wrong solution to the problem. Why are you using *that* web page as the be-all-end-all division of commands? > > > WBR > -- > Boris Samorodov (bsam) > Research Engineer, http://www.ipt.ru Telephone & Internet SP > FreeBSD committer, http://www.FreeBSD.org The Power To Serve > -- Derrick From bsam at ipt.ru Tue Dec 16 10:06:52 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Tue Dec 16 10:06:58 2008 Subject: openafs port(s) in general In-Reply-To: (Derrick Brashear's message of "Tue\, 16 Dec 2008 12\:56\:42 -0500") References: <37396849@bs1.sp34.ru> <4947B153.5060101@arcadebsd.org> <4947B7C4.8030004@arcadebsd.org> <35608451_-_@bb.ipt.ru> <69525564@bb.ipt.ru> <03443077@bb.ipt.ru> <37361713@bb.ipt.ru> Message-ID: <39127044@bb.ipt.ru> "Derrick Brashear" writes: > On Tue, Dec 16, 2008 at 11:49 AM, Boris Samorodov wrote: >> "Derrick Brashear" writes: >> >> > i could edit the web page in question if it'd help ;) >> >> Yes, please. > > Actually, it's the wrong solution to the problem. > > Why are you using *that* web page as the be-all-end-all division of > commands? Because there is no better solution. Do you have one? WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From shadow at gmail.com Tue Dec 16 10:17:51 2008 From: shadow at gmail.com (Derrick Brashear) Date: Tue Dec 16 10:17:58 2008 Subject: openafs port(s) in general In-Reply-To: <39127044@bb.ipt.ru> References: <37396849@bs1.sp34.ru> <4947B7C4.8030004@arcadebsd.org> <35608451_-_@bb.ipt.ru> <69525564@bb.ipt.ru> <03443077@bb.ipt.ru> <37361713@bb.ipt.ru> <39127044@bb.ipt.ru> Message-ID: On Tue, Dec 16, 2008 at 1:06 PM, Boris Samorodov wrote: > "Derrick Brashear" writes: > > On Tue, Dec 16, 2008 at 11:49 AM, Boris Samorodov wrote: > >> "Derrick Brashear" writes: > >> > >> > i could edit the web page in question if it'd help ;) > >> > >> Yes, please. > > > > Actually, it's the wrong solution to the problem. > > > > Why are you using *that* web page as the be-all-end-all division of > > commands? > > Because there is no better solution. Do you have one? > I gave you a list which got ignored, like 4 hours ago. Editing this web page to put the same list in it is any better why? From bsam at ipt.ru Tue Dec 16 10:44:41 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Tue Dec 16 10:44:47 2008 Subject: openafs port(s) in general In-Reply-To: (Derrick Brashear's message of "Tue\, 16 Dec 2008 13\:17\:50 -0500") References: <37396849@bs1.sp34.ru> <4947B7C4.8030004@arcadebsd.org> <35608451_-_@bb.ipt.ru> <69525564@bb.ipt.ru> <03443077@bb.ipt.ru> <37361713@bb.ipt.ru> <39127044@bb.ipt.ru> Message-ID: <57049240@bb.ipt.ru> "Derrick Brashear" writes: > On Tue, Dec 16, 2008 at 1:06 PM, Boris Samorodov wrote: >> "Derrick Brashear" writes: >> > On Tue, Dec 16, 2008 at 11:49 AM, Boris Samorodov wrote: >> >> "Derrick Brashear" writes: >> >> >> >> > i could edit the web page in question if it'd help ;) >> >> >> >> Yes, please. >> > >> > Actually, it's the wrong solution to the problem. >> > >> > Why are you using *that* web page as the be-all-end-all division of >> > commands? >> >> Because there is no better solution. Do you have one? > > I gave you a list which got ignored, like 4 hours ago. Why do you think your list got ignored? ;-) I've been working with it. Then that was an opinion of one person. Nobody voted for it. > Editing this web page > to put the same list in it is any better why? Your mail is your personal opinion. That site is (or am I mistaken?) an official place where to look for information. You know, I could have committed the (openafs) port since I got it build/install/deinstall succesfully. But I'd better don't do it. And discuss a better ways/decisions/etc. at this list. Because since the port gets committed it should reflect official position of those who *can* make decisions: developers, admins, power users. And because it will be (very) hard and/or painful to change the port (it's infrastructure, directory layout) later. One thing to say something while other to do it. Changes to that page (I think) should have been discussed and/or approved by some community. If those changes appear that may influence not only a FreeBSD port. WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From shadow at gmail.com Tue Dec 16 10:48:31 2008 From: shadow at gmail.com (Derrick Brashear) Date: Tue Dec 16 10:48:38 2008 Subject: openafs port(s) in general In-Reply-To: <57049240@bb.ipt.ru> References: <37396849@bs1.sp34.ru> <69525564@bb.ipt.ru> <03443077@bb.ipt.ru> <37361713@bb.ipt.ru> <39127044@bb.ipt.ru> <57049240@bb.ipt.ru> Message-ID: On Tue, Dec 16, 2008 at 1:44 PM, Boris Samorodov wrote: > "Derrick Brashear" writes: > > On Tue, Dec 16, 2008 at 1:06 PM, Boris Samorodov wrote: > >> "Derrick Brashear" writes: > >> > On Tue, Dec 16, 2008 at 11:49 AM, Boris Samorodov > wrote: > >> >> "Derrick Brashear" writes: > >> >> > >> >> > i could edit the web page in question if it'd help ;) > >> >> > >> >> Yes, please. > >> > > >> > Actually, it's the wrong solution to the problem. > >> > > >> > Why are you using *that* web page as the be-all-end-all division of > >> > commands? > >> > >> Because there is no better solution. Do you have one? > > > > I gave you a list which got ignored, like 4 hours ago. > > Why do you think your list got ignored? ;-) I've been working with it. > Then that was an opinion of one person. Nobody voted for it. > > > Editing this web page > > to put the same list in it is any better why? > > Your mail is your personal opinion. No. My mail reflects how OpenAFS works when you type "make install", which is at least as official as: > That site is (or am I mistaken?) > an official place where to look for information. It is. That I can edit the page (which, again, has nothing to do with where the commands get installed, but reflects only man subdirectories) suggests either my opinion has some bearing, or that I am clever and apply the policy OpenAFS has shaped to my work. I'd like to think it's the latter. > > You know, I could have committed the (openafs) port since I got it > build/install/deinstall succesfully. But I'd better don't do it. > And discuss a better ways/decisions/etc. at this list. Because > since the port gets committed it should reflect official position > of those who *can* make decisions: developers, admins, power users. > And because it will be (very) hard and/or painful to change the port > (it's infrastructure, directory layout) later. > I agree. Better to get it right, now, before people are depending on it. > > One thing to say something while other to do it. Changes to that page (I > think) should have been discussed and/or approved by some community. If > those changes appear that may influence not only a FreeBSD port. > See above. And let me try again to state what seems to be going over your head. That page has nothing to do with what the commands are for. It reflects which man pages are in man1 ("User commands") and which are in man8 ("Administrative commands") That's it. It's not policy, or any such other thing. From bsam at ipt.ru Tue Dec 16 12:26:37 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Tue Dec 16 12:26:44 2008 Subject: openafs port(s) in general In-Reply-To: (Derrick Brashear's message of "Tue\, 16 Dec 2008 13\:48\:30 -0500") References: <37396849@bs1.sp34.ru> <69525564@bb.ipt.ru> <03443077@bb.ipt.ru> <37361713@bb.ipt.ru> <39127044@bb.ipt.ru> <57049240@bb.ipt.ru> Message-ID: <57035445@bs1.sp34.ru> "Derrick Brashear" writes: > That page has nothing to do with what the commands are for. It reflects > which man pages are in man1 ("User commands") and which are in man8 > ("Administrative commands") > > That's it. It's not policy, or any such other thing. Hm, OK. I hoped that man pages reflect something meaningful. Thanks. WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From shadow at gmail.com Tue Dec 16 12:30:06 2008 From: shadow at gmail.com (Derrick Brashear) Date: Tue Dec 16 12:30:18 2008 Subject: openafs port(s) in general In-Reply-To: <57035445@bs1.sp34.ru> References: <37396849@bs1.sp34.ru> <03443077@bb.ipt.ru> <37361713@bb.ipt.ru> <39127044@bb.ipt.ru> <57049240@bb.ipt.ru> <57035445@bs1.sp34.ru> Message-ID: On Tue, Dec 16, 2008 at 3:26 PM, Boris Samorodov wrote: > "Derrick Brashear" writes: > > > That page has nothing to do with what the commands are for. It reflects > > which man pages are in man1 ("User commands") and which are in man8 > > ("Administrative commands") > > > > That's it. It's not policy, or any such other thing. > > Hm, OK. I hoped that man pages reflect something meaningful. They do. Command usage. Trying to read higher meaning from the location of them, less so. From gad at FreeBSD.org Tue Dec 16 13:46:00 2008 From: gad at FreeBSD.org (Garance A Drosehn) Date: Tue Dec 16 13:46:07 2008 Subject: openafs port(s) in general In-Reply-To: <57049240@bb.ipt.ru> References: <37396849@bs1.sp34.ru> <4947B7C4.8030004@arcadebsd.org> <35608451_-_@bb.ipt.ru> <69525564@bb.ipt.ru> <03443077@bb.ipt.ru> <37361713@bb.ipt.ru> <39127044@bb.ipt.ru> <57049240@bb.ipt.ru> Message-ID: At 9:44 PM +0300 12/16/08, Boris Samorodov wrote: >"Derrick Brashear" writes: > >> Editing this web page >> to put the same list in it is any better why? > >Your mail is your personal opinion. That site is (or am I mistaken?) >an official place where to look for information. > >You know, I could have committed the (openafs) port since I got it >build/install/deinstall succesfully. But I'd better don't do it. >And discuss a better ways/decisions/etc. at this list. Because >since the port gets committed it should reflect official position >of those who *can* make decisions: developers, admins, power users. >And because it will be (very) hard and/or painful to change the port >(it's infrastructure, directory layout) later. > >One thing to say something while other to do it. Changes to that >page (I think) should have been discussed and/or approved by some >community. If those changes appear that may influence not only a >FreeBSD port. Note that Derrick's opinion is much Much more experienced than the average "person on the street" when it comes to OpenAFS. -- Garance Alistair Drosehn = drosehn@rpi.edu Senior Systems Programmer or gad@FreeBSD.org Rensselaer Polytechnic Institute; Troy, NY; USA From bsam at ipt.ru Tue Dec 16 14:02:36 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Tue Dec 16 14:02:42 2008 Subject: asetkey (and aklog) is not build Message-ID: <90959685@bs1.sp34.ru> Hello List, I've just find out that asetkey (and aklog) is not build by the port. That seems to be a result of ignoring src/aklog. While an official package at OpenAFS site contains it at atdest/root.server/usr/afs/bin directory (aklog is at dest/bin). README says that aklog is not needed because heimdal afslog does (almost) the same thing. But nothing is said about asetkey. I'm not sure if I messed up my config. Can anybody confirm if the problem exists? Or maybe this file may be skipped (since it uses only kerberos v4)? WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From matt at linuxbox.com Tue Dec 16 14:14:39 2008 From: matt at linuxbox.com (Matt Benjamin) Date: Tue Dec 16 14:14:45 2008 Subject: asetkey (and aklog) is not build In-Reply-To: <90959685@bs1.sp34.ru> References: <90959685@bs1.sp34.ru> Message-ID: <49482843.1020304@linuxbox.com> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 Hi, The asetkey and aklog programs are part of the openafs client and server suite, respectively, which should be available. FWIW, one reason for these not building in ordinary openafs compiles (if that's an issue?) is failure to identify a working kerberos build environment (configure that includes - --with-krb5-config=/usr/bin/krb5-config would help, in that case). I apologize if that no longer makes sense in the port build framework, which I haven't experimented with, yet. Matt Boris Samorodov wrote: > Hello List, > > > I've just find out that asetkey (and aklog) is not build by the > port. That seems to be a result of ignoring src/aklog. While an > official package at OpenAFS site contains it at > atdest/root.server/usr/afs/bin directory (aklog is at dest/bin). > > README says that aklog is not needed because heimdal afslog does > (almost) the same thing. But nothing is said about asetkey. > > I'm not sure if I messed up my config. Can anybody confirm if the > problem exists? Or maybe this file may be skipped (since it uses > only kerberos v4)? > > > WBR - -- Matt Benjamin The Linux Box 206 South Fifth Ave. Suite 150 Ann Arbor, MI 48104 http://linuxbox.com tel. 734-761-4689 fax. 734-769-8938 cel. 734-216-5309 -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.7 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFJSChDJiSUUSaRdSURCBnGAJ42tIMC86CF/UrdK98Fw0uer+1ubgCgjnHq tKNs4Laexh8jnw66AGtCtfE= =/HUo -----END PGP SIGNATURE----- From bsam at ipt.ru Tue Dec 16 14:49:24 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Tue Dec 16 14:49:34 2008 Subject: [solved] Re: asetkey (and aklog) is not build In-Reply-To: <49482843.1020304@linuxbox.com> (Matt Benjamin's message of "Tue\, 16 Dec 2008 17\:14\:27 -0500") References: <90959685@bs1.sp34.ru> <49482843.1020304@linuxbox.com> Message-ID: <24876877_-_@bs1.sp34.ru> Matt Benjamin writes: > Hi, > > The asetkey and aklog programs are part of the openafs client and server > suite, respectively, which should be available. > > FWIW, one reason for these not building in ordinary openafs compiles (if > that's an issue?) Matt, you were right once more. That was the issue. > is failure to identify a working kerberos build > environment (configure that includes > --with-krb5-config=/usr/bin/krb5-config would help, in that case). I --with-krb5-conf=/usr/bin/krb5-config helped (mind option name). Great, thanks! > apologize if that no longer makes sense in the port build framework, > which I haven't experimented with, yet. Seems just fine imho. > Boris Samorodov wrote: >> Hello List, >> >> >> I've just find out that asetkey (and aklog) is not build by the >> port. That seems to be a result of ignoring src/aklog. While an >> official package at OpenAFS site contains it at >> atdest/root.server/usr/afs/bin directory (aklog is at dest/bin). >> >> README says that aklog is not needed because heimdal afslog does >> (almost) the same thing. But nothing is said about asetkey. >> >> I'm not sure if I messed up my config. Can anybody confirm if the >> problem exists? Or maybe this file may be skipped (since it uses >> only kerberos v4)? WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From bsam at ipt.ru Tue Dec 16 15:26:39 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Tue Dec 16 15:26:45 2008 Subject: openafs server executables In-Reply-To: (Derrick Brashear's message of "Tue\, 16 Dec 2008 08\:37\:49 -0500") References: <37396849@bs1.sp34.ru> <01705453@bs1.sp34.ru> <49453004.9010107@linuxbox.com> <83768316_-_@bs1.sp34.ru> Message-ID: <71276786@bs1.sp34.ru> Hello Derrick and All, "Derrick Brashear" writes: > Two copies of fs, vos, pts, bos, kas, asetkey, tokens, udebug, klog get > installed normally. you want copies of at least fs, vos, bos, asetkey, > udebug, pts (maybe not klog, tokens) in server, even if they are installed > elsewhere in client. I've taken the second list as a rule. Let me show what I managed to get. This is a list of executables. ===== Common binaries (should be installed always): ----- bin/ asetkey bos fs pts udebug sbin/ vos ----- Server binaries: ----- libexec/openafs/ buserver fileserver kaserver ptserver salvager upclient upserver vlserver volserver sbin/ aklog bosserver ka-forwarder kadb_check kdb kpwvalid prdb_check pt_util rmtsysd vldb_check vldb_convert voldump volinfo vsys ----- Client binaries: ----- bin/ afsmonitor cmdebug dlog dpass klog klog.krb knfs kpasswd livesys pagsh pagsh.krb scout sys tokens tokens.krb unlog up xstat_cm_test xstat_fs_test sbin/ afsd ----- ===== Changes/fixes/else? WBR -- bsam Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From minimarmot at gmail.com Tue Dec 16 15:35:22 2008 From: minimarmot at gmail.com (Ben Kaduk) Date: Tue Dec 16 15:35:29 2008 Subject: openafs server executables In-Reply-To: <71276786@bs1.sp34.ru> References: <37396849@bs1.sp34.ru> <01705453@bs1.sp34.ru> <49453004.9010107@linuxbox.com> <83768316_-_@bs1.sp34.ru> <71276786@bs1.sp34.ru> Message-ID: <47d0403c0812161535h565d360cx42827034eede500a@mail.gmail.com> On Tue, Dec 16, 2008 at 6:26 PM, Boris Samorodov wrote: > Hello Derrick and All, > > "Derrick Brashear" writes: > >> Two copies of fs, vos, pts, bos, kas, asetkey, tokens, udebug, klog get >> installed normally. you want copies of at least fs, vos, bos, asetkey, >> udebug, pts (maybe not klog, tokens) in server, even if they are installed >> elsewhere in client. > > I've taken the second list as a rule. Let me show what I managed to > get. This is a list of executables. > ===== > Common binaries (should be installed always): > ----- > bin/ > asetkey > bos > fs > pts > udebug > sbin/ > vos > ----- > > Server binaries: > ----- > libexec/openafs/ > buserver > fileserver > kaserver > ptserver > salvager > upclient > upserver > vlserver > volserver > sbin/ > aklog > bosserver > ka-forwarder > kadb_check > kdb > kpwvalid > prdb_check > pt_util > rmtsysd > vldb_check > vldb_convert > voldump > volinfo > vsys > ----- > > Client binaries: > ----- > bin/ > afsmonitor > cmdebug > dlog > dpass > klog > klog.krb > knfs > kpasswd > livesys > pagsh > pagsh.krb > scout > sys > tokens > tokens.krb > unlog > up > xstat_cm_test > xstat_fs_test > sbin/ > afsd > ----- > ===== > > Changes/fixes/else? I thought aklog was a client utility, but everything else looks okay to my (limited) experience. -Ben Kaduk From matt at linuxbox.com Tue Dec 16 15:47:42 2008 From: matt at linuxbox.com (Matt Benjamin) Date: Tue Dec 16 15:47:48 2008 Subject: openafs server executables In-Reply-To: <47d0403c0812161535h565d360cx42827034eede500a@mail.gmail.com> References: <37396849@bs1.sp34.ru> <01705453@bs1.sp34.ru> <49453004.9010107@linuxbox.com> <83768316_-_@bs1.sp34.ru> <71276786@bs1.sp34.ru> <47d0403c0812161535h565d360cx42827034eede500a@mail.gmail.com> Message-ID: <49483E11.7020202@linuxbox.com> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 aklog is a client utility. as far as I know, too, asetkey is only useful on server machines (add keys to KeyFile) Ben Kaduk wrote: u> > > I thought aklog was a client utility, but everything else looks okay > to my (limited) experience. > > -Ben Kaduk > _______________________________________________ > freebsd-afs@freebsd.org mailing list > http://lists.freebsd.org/mailman/listinfo/freebsd-afs > To unsubscribe, send any mail to "freebsd-afs-unsubscribe@freebsd.org" - -- Matt Benjamin The Linux Box 206 South Fifth Ave. Suite 150 Ann Arbor, MI 48104 http://linuxbox.com tel. 734-761-4689 fax. 734-769-8938 cel. 734-216-5309 -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.7 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFJSD4RJiSUUSaRdSURCHenAKCCtzk51Nc8d12ll/S6EL6fia9tHACbBxzD tuDbLmNMTrQzdd/8Qm5WAsE= =eijb -----END PGP SIGNATURE----- From matt at linuxbox.com Tue Dec 16 15:49:01 2008 From: matt at linuxbox.com (Matt Benjamin) Date: Tue Dec 16 15:49:07 2008 Subject: [solved] Re: asetkey (and aklog) is not build In-Reply-To: <24876877_-_@bs1.sp34.ru> References: <90959685@bs1.sp34.ru> <49482843.1020304@linuxbox.com> <24876877_-_@bs1.sp34.ru> Message-ID: <49483E60.5030502@linuxbox.com> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 I'm sorry, Ben, Boris was listening to me, probably, and I had these two reversed, below. Matt Boris Samorodov wrote: > Matt Benjamin writes: > >> Hi, >> >> The asetkey and aklog programs are part of the openafs client and server >> suite, respectively, which should be available. - -- Matt Benjamin The Linux Box 206 South Fifth Ave. Suite 150 Ann Arbor, MI 48104 http://linuxbox.com tel. 734-761-4689 fax. 734-769-8938 cel. 734-216-5309 -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.7 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFJSD5gJiSUUSaRdSURCHVyAKCNapNK0Em+zHEqPLfhpytEaHA67ACghSwP zKiiASTFpYC8ZZCrcoATQOM= =/U/A -----END PGP SIGNATURE----- From bsam at ipt.ru Wed Dec 17 00:02:54 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Wed Dec 17 00:03:00 2008 Subject: openafs server executables In-Reply-To: <47d0403c0812161535h565d360cx42827034eede500a@mail.gmail.com> (Ben Kaduk's message of "Tue\, 16 Dec 2008 18\:35\:22 -0500") References: <37396849@bs1.sp34.ru> <01705453@bs1.sp34.ru> <49453004.9010107@linuxbox.com> <83768316_-_@bs1.sp34.ru> <71276786@bs1.sp34.ru> <47d0403c0812161535h565d360cx42827034eede500a@mail.gmail.com> Message-ID: <92714739@bs1.sp34.ru> "Ben Kaduk" writes: > On Tue, Dec 16, 2008 at 6:26 PM, Boris Samorodov wrote: >> Hello Derrick and All, >> >> "Derrick Brashear" writes: >> >>> Two copies of fs, vos, pts, bos, kas, asetkey, tokens, udebug, klog get >>> installed normally. you want copies of at least fs, vos, bos, asetkey, >>> udebug, pts (maybe not klog, tokens) in server, even if they are installed >>> elsewhere in client. >> >> I've taken the second list as a rule. Let me show what I managed to >> get. This is a list of executables. [the list was here] >> Changes/fixes/else? > > I thought aklog was a client utility, but everything else looks okay > to my (limited) experience. Ben, Matt, I've got it, thanks. Here is the fixed list. ===== Common binaries (should be installed always): ----- bin/ asetkey bos fs pts udebug sbin/ vos ----- Server binaries: ----- libexec/openafs/ buserver fileserver kaserver ptserver salvager upclient upserver vlserver volserver sbin/ bosserver ka-forwarder kadb_check kdb kpwvalid prdb_check pt_util rmtsysd vldb_check vldb_convert voldump volinfo vsys ----- Client binaries: ----- bin/ afsmonitor aklog cmdebug dlog dpass klog klog.krb knfs kpasswd livesys pagsh pagsh.krb scout sys tokens tokens.krb unlog up xstat_cm_test xstat_fs_test sbin/ afsd ----- ===== WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From bsam at ipt.ru Wed Dec 17 01:49:37 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Wed Dec 17 01:49:43 2008 Subject: binary mtree for openafs port(s) In-Reply-To: <4947AFE8.3010506@arcadebsd.org> (elbarto@arcadebsd.org's message of "Tue\, 16 Dec 2008 14\:40\:56 +0100") References: <60571528@bs1.sp34.ru> <01693389@bb.ipt.ru> <4946E27E.1050803@arcadebsd.org> <49841540@bs1.sp34.ru> <4947AFE8.3010506@arcadebsd.org> Message-ID: <03437088@bb.ipt.ru> ElBarto writes: > Well, > For the client you only need all the binaries for mounting an afs > server (don't know exactly which ones). > This is the list I suggest : [list was here] > > $PREFIX/sbin/fstrace > $PREFIX/sbin/kpwvalid > $PREFIX/sbin/rxdebug I'm not sure either but it seems to me that those might belong to a server part. WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From matt at linuxbox.com Wed Dec 17 05:47:20 2008 From: matt at linuxbox.com (Matt Benjamin) Date: Wed Dec 17 05:47:27 2008 Subject: binary mtree for openafs port(s) In-Reply-To: <03437088@bb.ipt.ru> References: <60571528@bs1.sp34.ru> <01693389@bb.ipt.ru> <4946E27E.1050803@arcadebsd.org> <49841540@bs1.sp34.ru> <4947AFE8.3010506@arcadebsd.org> <03437088@bb.ipt.ru> Message-ID: <494902DA.6000303@linuxbox.com> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 Hi, Boris Samorodov wrote: >> $PREFIX/sbin/kpwvalid >> $PREFIX/sbin/rxdebug > > I'm not sure either but it seems to me that those might belong to > a server part. > > > WBR My impression is rxdebug -- used everywhere for troubleshooting, valuable to have on any afs host (both/base) fstrace -- like cmdebug, is used (infrequently, afaik) to debug client problems (client) kpwvalid -- support utility for kas kpasswd (KPWVALID(8)). Perhaps not critical what is done with this, but is part of kas/kaserver client infrastructure. (client) Matt - -- Matt Benjamin The Linux Box 206 South Fifth Ave. Suite 150 Ann Arbor, MI 48104 http://linuxbox.com tel. 734-761-4689 fax. 734-769-8938 cel. 734-216-5309 -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.7 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFJSQLaJiSUUSaRdSURCJdJAJ0RtEJZgPQDWZNtxR687WY2+jpIBQCcDTfS gRj8hs9nVjK8CKrxl3uqpBw= =yVop -----END PGP SIGNATURE----- From shadow at gmail.com Wed Dec 17 05:49:21 2008 From: shadow at gmail.com (Derrick Brashear) Date: Wed Dec 17 05:49:28 2008 Subject: openafs server executables In-Reply-To: <92714739@bs1.sp34.ru> References: <37396849@bs1.sp34.ru> <01705453@bs1.sp34.ru> <49453004.9010107@linuxbox.com> <83768316_-_@bs1.sp34.ru> <71276786@bs1.sp34.ru> <47d0403c0812161535h565d360cx42827034eede500a@mail.gmail.com> <92714739@bs1.sp34.ru> Message-ID: >> I thought aklog was a client utility, but everything else looks okay >> to my (limited) experience. > > Ben, Matt, I've got it, thanks. Here is the fixed list. > ===== > Common binaries (should be installed always): > ----- > bin/ > asetkey asetkey is server-only. it exists to create KeyFiles, which for security purposes you should *only* do on a server. > bos > fs > pts > udebug > sbin/ > vos > ----- > > Server binaries: > ----- > libexec/openafs/ > buserver > fileserver > kaserver > ptserver > salvager > upclient > upserver > vlserver > volserver > sbin/ > bosserver > ka-forwarder > kadb_check > kdb > kpwvalid > prdb_check > pt_util > rmtsysd > vldb_check > vldb_convert > voldump > volinfo > vsys > ----- > > Client binaries: > ----- > bin/ > afsmonitor > aklog > cmdebug > dlog > dpass > klog > klog.krb > knfs > kpasswd > livesys > pagsh > pagsh.krb > scout > sys > tokens > tokens.krb > unlog > up > xstat_cm_test > xstat_fs_test > sbin/ > afsd > ----- > ===== > > > WBR > -- > Boris Samorodov (bsam) > Research Engineer, http://www.ipt.ru Telephone & Internet SP > FreeBSD committer, http://www.FreeBSD.org The Power To Serve > _______________________________________________ > freebsd-afs@freebsd.org mailing list > http://lists.freebsd.org/mailman/listinfo/freebsd-afs > To unsubscribe, send any mail to "freebsd-afs-unsubscribe@freebsd.org" > -- Derrick From shadow at gmail.com Wed Dec 17 05:57:08 2008 From: shadow at gmail.com (Derrick Brashear) Date: Wed Dec 17 05:57:14 2008 Subject: binary mtree for openafs port(s) In-Reply-To: <03437088@bb.ipt.ru> References: <60571528@bs1.sp34.ru> <01693389@bb.ipt.ru> <4946E27E.1050803@arcadebsd.org> <49841540@bs1.sp34.ru> <4947AFE8.3010506@arcadebsd.org> <03437088@bb.ipt.ru> Message-ID: On Wed, Dec 17, 2008 at 4:49 AM, Boris Samorodov wrote: > ElBarto writes: > >> Well, >> For the client you only need all the binaries for mounting an afs >> server (don't know exactly which ones). >> This is the list I suggest : > [list was here] >> >> $PREFIX/sbin/fstrace >> $PREFIX/sbin/kpwvalid >> $PREFIX/sbin/rxdebug > > I'm not sure either but it seems to me that those might belong to > a server part. fstrace is client only, it debugs the client itself. rxdebug is universal. it can debug any rx service, client or server. kpwvalid is server only. fstrace, unlike cmdebug, must be run on the client. cmdebug, like rxdebug, can be *run* anywhere as it's a network probe. From matt at linuxbox.com Wed Dec 17 05:59:43 2008 From: matt at linuxbox.com (Matt Benjamin) Date: Wed Dec 17 05:59:50 2008 Subject: binary mtree for openafs port(s) In-Reply-To: References: <60571528@bs1.sp34.ru> <01693389@bb.ipt.ru> <4946E27E.1050803@arcadebsd.org> <49841540@bs1.sp34.ru> <4947AFE8.3010506@arcadebsd.org> <03437088@bb.ipt.ru> Message-ID: <494905C3.20308@linuxbox.com> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 I shouldn't have commented on that one. I've never run a kaserver. Matt Derrick Brashear wrote: kpwvalid is > server only - -- Matt Benjamin The Linux Box 206 South Fifth Ave. Suite 150 Ann Arbor, MI 48104 http://linuxbox.com tel. 734-761-4689 fax. 734-769-8938 cel. 734-216-5309 -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.7 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFJSQXDJiSUUSaRdSURCEHnAJ4totifuqPVQsDZxNgnvwJtXVTx3wCfRDWk 4Zfjs/uVR1hBDja41IXm/z4= =KRcO -----END PGP SIGNATURE----- From bsam at ipt.ru Wed Dec 17 06:16:24 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Wed Dec 17 06:16:31 2008 Subject: binary mtree for openafs port(s) In-Reply-To: <494902DA.6000303@linuxbox.com> (Matt Benjamin's message of "Wed\, 17 Dec 2008 08\:47\:06 -0500") References: <60571528@bs1.sp34.ru> <01693389@bb.ipt.ru> <4946E27E.1050803@arcadebsd.org> <49841540@bs1.sp34.ru> <4947AFE8.3010506@arcadebsd.org> <03437088@bb.ipt.ru> <494902DA.6000303@linuxbox.com> Message-ID: <05180009@bb.ipt.ru> Matt Benjamin writes: > Boris Samorodov wrote: >>> $PREFIX/sbin/kpwvalid >>> $PREFIX/sbin/rxdebug >> >> I'm not sure either but it seems to me that those might belong to >> a server part. >> >> >> WBR > > My impression is > > rxdebug -- used everywhere for troubleshooting, valuable to have on any > afs host (both/base) Added to common_sbin_file_list. > fstrace -- like cmdebug, is used (infrequently, afaik) to debug client > problems (client) > > kpwvalid -- support utility for kas kpasswd (KPWVALID(8)). Perhaps not > critical what is done with this, but is part of kas/kaserver client > infrastructure. (client) Both added to client_sbin_file_list. Thanks! All files were added to sbin directory since by default those files are installed to sbin directory. WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From bsam at ipt.ru Wed Dec 17 06:17:22 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Wed Dec 17 06:17:28 2008 Subject: openafs server executables In-Reply-To: (Derrick Brashear's message of "Wed\, 17 Dec 2008 08\:49\:20 -0500") References: <37396849@bs1.sp34.ru> <01705453@bs1.sp34.ru> <49453004.9010107@linuxbox.com> <83768316_-_@bs1.sp34.ru> <71276786@bs1.sp34.ru> <47d0403c0812161535h565d360cx42827034eede500a@mail.gmail.com> <92714739@bs1.sp34.ru> Message-ID: <39109951@bb.ipt.ru> "Derrick Brashear" writes: >> bin/ >> asetkey > > asetkey is server-only. it exists to create KeyFiles, which for > security purposes you should *only* do on a server. Got it, thanks! WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From bsam at ipt.ru Wed Dec 17 06:30:00 2008 From: bsam at ipt.ru (Boris Samorodov) Date: Wed Dec 17 06:30:09 2008 Subject: binary mtree for openafs port(s) In-Reply-To: (Derrick Brashear's message of "Wed\, 17 Dec 2008 08\:57\:06 -0500") References: <60571528@bs1.sp34.ru> <01693389@bb.ipt.ru> <4946E27E.1050803@arcadebsd.org> <49841540@bs1.sp34.ru> <4947AFE8.3010506@arcadebsd.org> <03437088@bb.ipt.ru> Message-ID: <57023658@bb.ipt.ru> "Derrick Brashear" writes: > On Wed, Dec 17, 2008 at 4:49 AM, Boris Samorodov wrote: >> ElBarto writes: >> >>> Well, >>> For the client you only need all the binaries for mounting an afs >>> server (don't know exactly which ones). >>> This is the list I suggest : >> [list was here] >>> >>> $PREFIX/sbin/fstrace >>> $PREFIX/sbin/kpwvalid >>> $PREFIX/sbin/rxdebug >> >> I'm not sure either but it seems to me that those might belong to >> a server part. > > fstrace is client only, it debugs the client itself. rxdebug is > universal. it can debug any rx service, client or server. kpwvalid is > server only. > fstrace, unlike cmdebug, must be run on the client. cmdebug, like > rxdebug, can be *run* anywhere as it's a network probe. OK, thank you for your help! So far I've written a script which installs all files to apropriate locations. There are some minor questions left, but this part of a port (binary mtree) seems to be done soon. WBR -- Boris Samorodov (bsam) Research Engineer, http://www.ipt.ru Telephone & Internet SP FreeBSD committer, http://www.FreeBSD.org The Power To Serve From minimarmot at gmail.com Wed Dec 17 11:51:25 2008 From: minimarmot at gmail.com (Ben Kaduk) Date: Wed Dec 17 11:51:32 2008 Subject: Changes to OpenAFS to support FreeBSD 8-CURRENT In-Reply-To: <494475C2.7000303@linuxbox.com> References: <47d0403c0812131824i489c50c6w456e62e43ecabdeb@mail.gmail.com> <494475C2.7000303@linuxbox.com> Message-ID: <47d0403c0812171151s5153b68qf460d19d7a66687f@mail.gmail.com> On Sat, Dec 13, 2008 at 9:56 PM, Matt Benjamin wrote: > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA256 > > Hi Ben, > > I got your previous email, but wasn't sure I could get to this. I > haven't in fact been focused on 8-CURRENT, but on vnode locking and > refcounting correctness across 7.0 to 8-CURRENT. (Making headway...) > > Anyway, I set that aside and whacked through a similar effort. The > things you got sound right to me, though I'm surprised krx and > afs_server.c compiled, but, it might have to do with whether you have > network stack virtualization enabled. The module doesn't load > presumably because of unresolved externals--here either the > aforementioned (like in_ifaddrhead), and probably "suser(thread)" which > is gone, replaced with "priv_check(thread)" (what makes me think that > one won't be around long). > > Openafs-1.4.8 (openafs.org) with RT # 123933 compiles and loads, as > above, but I have not run the code, it's just to block something in, ATM. > It's possible that I'm doing something wrong, but I'm getting panics with returning from a syscall with a lock held. I have a project that I need to finish up for my classes, here, so I shouldn't spend more time on this for a few days. If anyone really wants a core before then, let me know, but I'm pretty sure that my libafs.ko doesn't have debugging symbols, so it's probably not worth it. Just FYI, Ben Kaduk