(Illustration by Gaich Muramatsu)
On Thu, Nov 29, 2001 at 11:00:17AM +0100, Andrea Cerrito wrote: > So I tried to restart venus on both cluster node: on cluster A were > no problems, but on cluster B I found this on console log (and venus > never mounted /coda). > > 10:53:05 6369 cache files on free-list > Assertion failed: vol->IsReplicated(), file "fso1.cc", line 1889 > Sleeping forever. You may use gdb to attach to process 4613. > > My problem is: > > - - why venus can't mount /coda? Because during the startup venus tried to do a CML optimization on an object that logically can't have a CML because it is in a non-replicated volume. This seems to be an artifact of the local-global repair code that is 'temporarily' moving object to a fake non-replicated volume to avoid conflicts with the global replica. However it doesn't always clean up nicely after itself when no repair is done, or when repair fails. > - - how can I force /coda to be mounted to retrieve data? I'm sure > there are data into client cache, and I must retrieve it. I do not know of a way to nicely recover from this. It pretty much requires a client reinitialization, but you should still be able to recover any write-disconnected modifications ('lost' files) from /usr/coda/spool/<userid>/<volume>.tar (or /var/lib/coda/spool on Debian) JanReceived on 2001-11-30 13:24:59