On Tue, Aug 28, 2001 at 08:16:43AM +0800, Jeremy Malcolm wrote:
> Jan Harkes wrote:
> > 
> > Are there any errors in /vice/srv/SrvLog about volumes that cannot be
> > found?
> 
> Yes, I have now found this section of the logs as a possible problem
> (there is something odd about the logs because they don't write fully
> until I stop the server):

Well, there was this fsync that was heavily chewing up CPU time in
kernel space when I was profiling a server. That fsync happens to flush
the log to disk, I'll put it back in the next release.

> 22:28:13 client_GetVenusId: got new host 192.168.0.1:1282
> 22:28:13 Building callback conn.
> 22:28:13 No idle WriteBack conns, building new one
> 22:28:13 Writeback message to 192.168.0.1 port 1282 on conn 34b5acb8
> succeeded
> 22:28:13 RevokeWBPermit on conn 34b5acb8 returned 0
> 22:28:13 GetVolObj: VGetVolume(7f000000) error 103
> 22:28:13 GrabFsObj, GetVolObj error Volume not online

Volume not online? Something must have gone wrong during salvaging,
a terminally ill volume is forced offline when the salvager cannot
be sure the state is valid.

> VRDB and VLDB are identical on both servers.  7f000000 is my
> non-replicated root volume on servalan.  I get this error multiple
> times.  Actually, I don't *care* about 7f000000.  Can I just get rid of
> it and only have the replicated volumes?

Sure, just change /vice/db/ROOTVOLUME to have the name of one of the
replicated volumes, or create a new replicated rootvolume. Clients that
have some files from this volume in their cache will be complaining a bit
louder until they've managed to "uncache" those objects.

Jan