[Dovecot] Dovecot+NFS: Copying messages causes failures (possible SIGABRT)

[Tony Kay]

Environment:

Dovecot 1.0beta8
Host: 64 bit Red Hat AS4 Linux (2.6.9-34.0.1)
Disks: NFS netapp

In the case below, the two connections were going through two separate servers sharing disks via an NFS NetApp.

Here is the sequence of operations:

1. Open a connection to a folder (say mail/Trash)
2. Open a connection to another folder (say Inbox)
3. Move messages from Inbox to Trash using this sequence on Inbox:

   uid copy #,# "mail/Trash"
   uid store #,# +flags (\Deleted)
   check
   expunge

4. Next, access open connection to Trash:

   noop (completes OK, but does not see new messages)
...wait a few seconds...
   noop (connection drops...no response)

Attempt to reconnect: mailbox locked. Will override in 270 seconds.

The dovecot logs contain the following for the exact time when the connection drops:

dovecot: IMAP(tkay): file mail-index-sync-update.c: line 686 (mail_index_sync_update_index): assertion failed: (view->hdr.messages_count == map->hdr.messages_count)

I cannot correlate this to the user/folder, but this message is usually nearby:

child 22915 (imap) killed with signal 6
  
--
Anthony Kay
University Computing Center
(541) 346-1719
GPG Fingerprint: B0DB D46A 60AF FAE7 A94A  5075 0CB4 4D88 9F4F 7F09

  Perfection uncanned 
Like a beautiful redhead
 Fresh from her trailer

        from Spam-Ku : Tranquil Reflections on Luncheon Loaf
             by John Cho