src upgrade from 4.11-p16 to 5.4 possible?

[Jerry McAllister]

> 
> Hello all,
> 
> I am running a set of 5 FreeBSD 4.11 p16 servers here at the library.
> 
> For various reasons we are unable to 'get new drives' and install 
> FreeBSD 5/6 and then migrate userland data to a new drive.
> 
> I have been able to get my hands on a test box from someone to test src 
> upgrading from 4.11 to something more current like 5.x and then updating 
> from there.

Trying to do an upgrade over major versions can uncover unexpected
problems.   I don't notice anything in your specs that would make 
it not work with the latest FreeBSD (though I am not up on the latest
CPU quirks).

Since you have an "extra" machine to work with, why don't you just do a
fresh install of FreeBSD 6.xxx on it and then build on that.  Install
the latest versions of the ports you are using on your servers and check
it all out and then move your userland from one of the 5 in-service systems 
on to it.  Then, replace that machine in-service with the test machine
and use the replaced machine to build the next one, etc.

You would get the newer UFS2 file system improvements which you 
would miss with just an upgrade and it would probably result in an
over-all cleaner system.  Everything should work just hunkie dorie.

If the test box is of too little capacity for normal service, just
continue the ripple install all the way around until the test box 
is the one pulled back out of service.   If you have done a careful
job of doing the first install and build on the test box you will 
have the necessary steps down and can probably do all 5, one at a 
time in a day.

////jerry

> 
> My standard running machines are Dual 933 Intel machines, 1G ram, scsi 
> drives.. etc
> 
> This test box is an AMD Mobile Sempron 2600+ w/ 512M ram, 80G ide drives.
> 
> http://www.freebsd.org/releases/5.4R/migration-guide.html is the guide 
> that I am following.. and I have done what I think are the right steps.. 
> and I am getting a crash/failure in a most unusual place.
> 
> Stop in /usr/src/usr.sbin/pcvt/vgaio.
> *** Error code 1
> 
> 
> ===> usr.sbin/pcvt/ispcvt
> cc -O -pipe   -c /usr/src/usr.sbin/pcvt/ispcvt/ispcvt.c
> cc -O -pipe    -o ispcvt ispcvt.o
> gzip -cn /usr/src/usr.sbin/pcvt/ispcvt/ispcvt.8 > ispcvt.8.gz
> ===> usr.sbin/pcvt/vgaio
> cc -O -pipe  -I/usr/obj/usr/src/usr.sbin/pcvt/vgaio 
> -I/usr/src/usr.sbin/pcvt/vgaio  -c vgaio.c
> In file included from /usr/src/usr.sbin/pcvt/vgaio/vgaio.y:56:
> /usr/obj/usr/src/i386/usr/include/machine/cpufunc.h:60: error: syntax 
> error before "bsfl"
> /usr/obj/usr/src/i386/usr/include/machine/cpufunc.h:60: error: syntax 
> error before "mask"
> /usr/obj/usr/src/i386/usr/include/machine/cpufunc.h: In function `bsfl':
> ... etc
> 
> The strange thing is this:
> CPU: Mobile AMD Sempron(tm) Processor 2600+ (799.92-MHz 686-class CPU)
> 
> this is what dmesg tells me about this cpu.. again this is under 4.11
> 
> So I am wondering if these things are related, and if I try this on an 
> Intel cpu will I still have this issue?
> 
> As my cvsup to RELENG_5_4 and then make buildworld fails at this point.
> 
> I have script'd the make buildworld which is a 9M text file, and it 
> compresses down to a ~500k bz/gz file if anyone is interested in it.
> 
> I have tried in single user mode as well again no difference.  I have 
> tried make -j4 buildworld (only gets there faster.. )
> 
> Thanks in advance any help is greatly appreciated.
> 
> (I'm going to try 5.3 and then if that doesn't work then I'll try 5.2.1)
> 
> - Brian
> _______________________________________________
> freebsd-questions at freebsd.org mailing list
> http://lists.freebsd.org/mailman/listinfo/freebsd-questions
> To unsubscribe, send any mail to "freebsd-questions-unsubscribe at freebsd.org"
>