On Wed, Jun 04, 2003 at 12:36:25AM -0700, Steve Simitzis wrote:
> i upgraded both my backup server and my coda server to 6.0.1. after doing
> this, the backup failed, crashing my coda server in the process:

Hmm, so that was 6.0.0 breakage. I had it myself, but I thought it was a
problem between 5.3.19 and 5.3.20. It happened when I went from 5.3.19
to a snapshot of the CVS tree around January.

> 00:17:13 VAttachVolumeById: vol 1000016 (sg.media.site.0.backup) attached and online
> 00:17:13 S_VolMakeBackups: backup (1000016) made of volume 1000006 
> 00:17:13 GrowVnodes: growing Small list from 47104 to 47616 for volume 0x1000017
> (and then it dies)
...
> Assertion failed: SRV_RVM(VolumeList[rwIndex]).data.nsmallLists == SRV_RVM(VolumeList[backupIndex]).data.nsmallLists, file "/usr/src/redhat/BUILD/coda-6.0.1/coda-src/volutil/vol-backup.cc", line 426
> EXITING! Bye!

Yeah, that's the one. As you discovered, purging the backup volumes
fixes it. And I haven't had a problem since. It is clearly a bug, and I
don't really know what caused it. It looks like the volume vnode list
in the original replica is smaller than that of the backup volume and
the backup volumes assume that it can only be bigger, and it will only
grow it's own list to a larger size.

Jan