New knob for ignoring readonly fss in 340.noid and 310.locate?

[Andre Albsmeier]

On Mon, 02-May-2011 at 09:04:37 +0200, Jason Hellenthal wrote:
> 
> Andre,
> 
> 
> On Mon, May 02, 2011 at 07:27:39AM +0200, Andre Albsmeier wrote:
> >On Mon, 02-May-2011 at 04:59:42 +0200, Jason Hellenthal wrote:
> >> 
> >> Andre,
> >> 
> >> 
> >> On Sun, May 01, 2011 at 10:19:30AM +0200, Andre Albsmeier wrote:
> >> >On Sat, 30-Apr-2011 at 23:31:57 +0200, Jason Hellenthal wrote:
> >> >> 
> >> >> By default snapshots directories are hidden and treated as a virtual
> >> >
> >> >Is it possible to hide snapshots directories in UFS?
> >> >
> >> 
> >> Snapshot directories on UFS are treated differently than they are in
> >> ZFS. UFS snapshot directories live as the base of the filesystem and are
> >> not auto-mounted perse when you cd(1) into them so therefore there isn't a
> >> need to hide them because they cannot be traversed.
> >
> >They are mounted and they have to be mounted (at least here). If
> >they weren't mounted, people couldn't access them. That's why
> >they are also being traversed by 310.locate and 340.noid. To
> >summarise:
> >
> >- I use UFS.
> >- My snapshots must be mounted.
> >- They are being traversed by 310.locate and 340.noid.
> >- I don't want the latter.
> >
> >To accomplish this, I can play around with (directory name dependent)
> >exclusion lists for 310.locate and 340.noid. I could also implement
> >a rdonly knob.
> >
> 
> Well for the case of 310.locate this is already of no concern since it
> can be finely tuned via /etc/locate.rc

I know. But as outlined in my initial mail, I would have to 
tune /etc/locate.rc for each directory by name on each machine.
A simple "ignore readonly mountpoints" knob would simplify
things a lot.

	-Andre

> 
> As for 340.noid I played around with that. It should have been modeled
> after 100.setuid but it wasn't. Also doesn't really provide any real
> useful information other than a filename on output.
> 
> Attached is an adjustment to 340.noid that I think you will like and
> here is some sample output.
> 
> $ sudo sh 340.noid
> 
> Check for files with an unknown user or group:
> Directories, /
> -rw-r--r--  1 404   404    0 May  2 02:44 /test2
> -rw-r--r--  1 root  404    0 May  2 02:52 /test3
> -rw-r--r--  1 404   wheel  0 May  2 02:53 /test4
> 
> After I finish this up I am going to propose a patch for both of
> 100.setuid and 340.noid so they can provide similiar output to the
> above.
> 
> For now I have just attached the modified script as it has to unset the
> weekly_noid_dir variable before it brings in your local periodic configs
> otherwise it wouldn't have let the script function with your config.
> 
> Ill write back with an update once that is finished.
> 
> PS: You should be able to just drop the attachment into the following
> directories for it to take effect and then adjust the directories via
> weekly_noid_dir per your periodic.conf or periodic.conf.local
> 
> /usr/src/etc/periodic/weekly
> /etc/periodic/weekly
> 
> -- 
> 
>  Regards, (jhell)
>  Jason Hellenthal
> 

> #!/bin/sh -
> #
> # $FreeBSD: stable/8/etc/periodic/weekly/340.noid 220107 2011-03-28 19:22:55Z dougb $
> #
> 
> # If there is a global system configuration file, suck it in.
> #
> if [ -r /etc/defaults/periodic.conf ]
> then
> 	. /etc/defaults/periodic.conf
> 	unset weekly_noid_dirs
> 	source_periodic_confs
> fi
> 
> case "$weekly_noid_enable" in
> 	[Yy][Ee][Ss])
> 		echo ""
> 		echo "Check for files with an unknown user or group:"
> 
> 		MP="${weekly_noid_dirs:-`mount -t ufs,zfs |awk '{print $3}' |xargs`}"
> 		echo "Directories, $MP"
> 		find -sxH $MP /dev/null \
> 			\( ! -fstype local -prune -or -name \* \) -and \
> 			\( -nogroup -o -nouser \) |sed 's/^/  /' |\
> 			xargs ls -l
> 		rc=$?
> 		;;
> 	*)  rc=0;;
> esac
> 
> exit $rc




-- 
Division by zero error -- multiplying by zero to recover...