On Tue, Jul 17, 2001 at 11:01:55PM +0800, tin wrote:
> 1.May i know why i have to doing this? 
> 
> You might have to edit vicetab on the SCM to also
> include,
> <new_server_name> /vicepa ftree width=64,depth=3

I believe that in the default configuration the /vice/db/vicetab file is
shared amongst all servers (i.e. replicated by updateclnt/updatesrv).

When a server starts up, it isn't going to just go scribble in random
places on your harddisks. So we have to tell the server where it can
store filedata. The 'exported' partitions are all in the vicetab file,
but since the non-SCM server setup doesn't run on the SCM and thus
cannot modify the shared configuration files (servers, VSGDB, vicetab)
there are still several things that have to be done by hand.

> and 'touch /vicepa/FTREEDB' on the new server to allow it to recognise
> and start using the /vicepa partition.

Second precaution. If you ever forget to mount /vicepa during bootup (or
the mount fails for some reason) the server at least won't start filling
up your rootpartition, even when the vicetab says it's OK to use /vicepa.

> 2. Create a new server entry mean create entry in /vice/db/servers  ?
> 3. new VSGDB entries mean create entry in /vice/db/VSGDB ?
> 3. , and create new volumes for the new VSG.  How ?
> 4. can you show me step by step ? (still confuse) :(

On the SCM,

# echo "newserver 71 >> /vice/db/servers
# echo "E0002001 SCM newserver" >> /vice/db/VSGDB
# echo "newserver /vicepa ftreedb width=64,depth=3" >> /vice/db/vicetab

Restart all codasrv processes.
    # /etc/rc.d/init.d/codasrv restart (RedHat)

or, # /etc/init.d/coda-server restart  (Debian)

or, # volutil shutdown
    # startserver &

or, # killall -9 codasrv
    # startserver

or, ctrl-alt-del, reboot.
    
Then when everything came back up check the SrvLog on the new server to
see whether it is using the /vicepa partition. Something like,

Partition /vicepa: inodes in use: 87207, total: 262144.
09:38:39 Partition /vicepa: 1930382K available (minfree=5%), 29298K free.

Then start creating the replicated volumes (from the SCM) and mount them
into the /coda tree from some client.

    SCM   # createvol_rep replicated_vol E0002001
    client$ cfs mkm /coda/newvol replicated_vol

Jan