Jan Harkes wrote:
> 
> If you recreated the volume using the createvol_rep script the
> bldvldb.sh and volutil makevrdb steps should already have been done for
> you.

Yes I'm using createvol_rep...

> What version of Coda are you using? Here bldvldb.sh start up with
> "Fetching volume lists from servers:".

This a Debian package of Coda 5.3.1.

I look at sources of bldvldb.sh, and it is using 
  updatefetch -h ${server} -r /vice/vol/VolumeList -l
${REMOTE}/${server}.list.new
instead of
  volutil getvolumelist ...

> This part was the 'pull', where the SCM collects all information from
> the machines in the Coda cell. The second part is where the updateclnt
> on the other servers discover that their versions of /vice/db/VRDB and
> /vice/db/VLDB are outdated.
> 
> Check whether updatesrv is running on the SCM, and the updateclnt
> daemons are running on all servers (including the SCM).

Everything seems OK.
Can it be possible that update agents died or segfault without prompting
me, and keep running in a sleeping loop, like codasrv do.

Thanks for all.

-Emmanuel