cvs pserver sig11 on 4.8-R

[Scott Mitchell]

On Sat, Jul 26, 2003 at 01:55:27PM +0200, arno at heho.snv.jussieu.fr wrote:
> Scott Mitchell <scott+freebsd at fishballoon.org> writes:
> 
> > On Wed, Jul 23, 2003 at 02:02:55PM -0700, Doug White wrote:
> > > 
> > > Does the new machine have ECC memory? If not, it could be something as
> > > simple as bad RAM in the new system.
> > 
> > I've just done four successive buildworlds on this machine, one with -j4,
> > just for fun.  They all completed without complaint, so I'm reasonably
> > confident there's no dodgy RAM here.
> 
> I second your expereinces : I see the same on my laboratory
> main server (Athlon 700) ever since cvs was updated from 1.10
> to 1.11 (4.7-R ?) :

I've just had a look in the (freebsd.org) cvs logs -- it looks like FreeBSD was
using cvs 1.11.1p1 for 4.5-R through 4.7-R, then moved to 1.11.2 and then
1.11.5 for 4.8-R.   I have not seen this problem with any release before 4.8,
although we skipped 4.7 entirely.  That said, without knowing what the bug is,
or even if there really is a bug, I'm not making any statements about which
versions it appears in :-) 

In the dodgy RAM department, the diagnostic tools for the server board do seem
to include a comprehensive RAM test, so I plan to give this and memtest86 a
workout when I can schedule some downtime on the server.  Will report back
when that is done.

Cheers,

	Scott