Hi,
I have 2 * 2.5.7 servers. The first is setup to replicate to the other
(which is currently empty). The 'master' has quite a bit of mail on it
(~90Gb).
As it's quite a lot - I brought up the replica, but left 'rolling'
replication off.
On the 'master' I then ran:
/usr/local/cyrus/bin/sync_client -v -A
This started chugging away through all the mailboxes - but then failed:
Apr 20 16:24:59 syncserver[17103]: twoskip: checkpointed
/vol/data/imap/mailboxes.db (304 records, 60352 => 56864 bytes) in 0.016
seconds
Apr 20 16:25:03 syncserver[17103]: twoskip: checkpointed
/vol/data/imap/user/m/mailbox.seen (70 records, 18384 => 16912 bytes) in
0.000 seconds
Apr 20 16:25:04 syncserver[17103]: IOERROR: failed to parse
/vol/data/imap/spool/sync./17103/9e4cb1ca93123a15dc49d84a9c7a9a60841a1079
Apr 20 16:25:04 syncserver[17103]: IOERROR: failed to append file
user.mailbox.CantProcess 1
Apr 20 16:25:04 master[17097]: process type:SERVICE name:syncserver
path:/usr/local/cyrus/bin/sync_server age:3096.847s pid:17103 signaled to
death by signal 6 (Abort trap)
Apr 20 16:25:04 kernel: Failed to write core file for process sync_server
(error 14)
Apr 20 16:25:04 kernel: pid 17103 (sync_server), uid 60: exited on signal 6
If I re-run the 'sync_client' command as above, it fails again at the same
point (just after starting out).
The directory in question ('/vol/data/imap/spool/sync./17103') has
1,517,996 files in it - but none are called
'9e4cb1ca93123a15dc49d84a9c7a9a60841a1079'.
Can I safely remove all of '/sync./*' and start the replication again? -
Anything else I can do to either fix / troubleshoot this?
As far as I can see "user.mailbox.CantProcess 1" on the 'master' doesn't
exist (i.e. there is no '1.' message file on the underlying file system).
Nor, on the replica. The mailbox looks "mostly" replicated - the next
message up (if it's doing them in numerical order) would be the file '27.'
- and that file looks unremarkable.
Thanks,
-Karl
