On Tue, Jul 25, 2000 at 12:45:54PM +0200, Florin Grad wrote:
> 
> Hello again,
> Thank you very much for your previous answers...
> 
> but I still have some more questions:
> 
> right after the installation I created a codaroot volume with
> creatvol_rep codaroot E0000100 /vicepa then I was asked for the backup,
> etc.
> 
> So this created a read/write replicated volume which resides only on the
> scm. (Well, this is already strange to have a replicated volume on a
> single server). But in this case, if you turn off the scm, the clients
> complain: "cannot find Root Volume" or something like that

Well, a singly replicated volume is a completely different beast
compared to a single volume as it allows for disconnected operation.
> 
> But, according to the docs, and I agree with that, it's better to have a
> readonly replicated rootvolume, so the non-scm servers can replace the scm
> one if it crashes or if it's off.

No, readonly replicated volumes are definitely deprecated, they miss
some vital information to allow for disconnected operation. The best
solution is to create a replicated volume, and modify
/vice/db/ROOTVOLUME to contain the name of the new rootvolume. The
clients will complain a bit on the next startup, but shouldn't have a
problem with it.

> Here is my error message
> 
> [root_at_sunlight coda]# volutil clone E0000100
> V_BindToServer: binding to host sunlight
> VolClone failed with Unknown RPC2 return code 103

The volume wasn't locked, but I really wouldn't spend any time with r/o
replication, it will be gone from the code at some point.

> Should one create a non-replicated root volume with volutil create ?
> The createvol command the docs are talking about doesn't even exist
> anymore...

Yes, non-replicated volumes are still there, in the form of
backupvolumes, and the underlying replicas of a replicated volume, but
since there is no possibility for cache revalidation, they can only be
used while fully connected to the servers.

> The right command will be volutil create /vicepa codaroot ?
> But in this case how do we specify E0000100 ?

E0000100 is the replication group, and therefore only used when creating
replicated volumes. Replicated volume information is stored in the VRDB,
while volumes replicas, clones, and backup volumes are only present in
the VLDB.

Jan