[Dovecot] Dovecot "assertion failed" errors
Dear all,
We are running Dovecot (on debian etch, Version 1.0.rc15-2) on 4
Loadbalanced servers using a NFS share for the homedirs. The
loadbalancer is using persistant connections (timeout 30min).
Lately, IMAP Users (mostly using Thunderbird) reported the following
errors:
- When trying to mark messages as read, nothing happened, message
was still displayed as unread. - When moving a Directory the client loads for a long time and then
hangs. Same was reported to happened when a user tried to open a mail.
When checking the logs at the time users reported those errors we
could always find one fo the following "assertion failed" errors:
file mail-index-sync-update.c: line 730
(mail_index_sync_update_index): assertion failed: (view-
hdr.messages_count == map->hdr.messages_count)
file mail-index.c: line 415 (mail_index_map_parse_keywords):
assertion failed: (ext->hdr_offset < map->hdr.header_size)
file mail-index.c: line 909 (mail_index_sync_from_transactions):
assertion failed: (prev_seq <= max_seq && (prev_seq != max_seq ||
prev_offset <= max_offset))
file mail-transaction-log-view.c: line 158
(mail_transaction_log_view_set): assertion failed: (min_file_seq !=
max_file_seq || min_file_offset <= max_file_offset)
file mail-transaction-log-view.c: line 229
(mail_transaction_log_view_set): assertion failed: (view->cur_offset
<= view->cur->sync_offset)
file message-body-search.c: line 393 (message_body_search_ctx):
assertion failed: (input->v_offset <= part->physical_pos)
I wonder if anyone has experienced similar problems and if there is a
solution to this strange behaviour.
University of Bern IT Department Central Systems Group
Christian Friedli E-Mail Services
Gesellschaftsstr. 6 CH-3012 Bern Switzerland Tel. +41 (0)31 631 38 38 Fax +41 (0)31 631 43 57 mailto:christian.friedli@id.unibe.ch http://www.mail.unibe.ch
--
Christian Friedli wrote:
Dear all,
We are running Dovecot (on debian etch, Version 1.0.rc15-2) on 4 Loadbalanced servers using a NFS share for the homedirs. The loadbalancer is using persistant connections (timeout 30min).
Upgrade...
Many, many bug fixes since the 1.0 release.
--
Best regards,
Charles
Dear Chales,
We made a upgrade to dovecot 1.0.0-1. It doesn't seem to help. We have the same problems whit this version. Any other ideas?
Best regards, Christian
Am 25.04.2007 um 19:20 schrieb Charles Marcus:
Christian Friedli wrote:
Dear all, We are running Dovecot (on debian etch, Version 1.0.rc15-2) on 4
Loadbalanced servers using a NFS share for the homedirs. The
loadbalancer is using persistant connections (timeout 30min).Upgrade...
Many, many bug fixes since the 1.0 release.
--
Best regards,
Charles
Hi Christian,
Please don't top-post...
We are running Dovecot (on debian etch, Version 1.0.rc15-2) on 4 Loadbalanced servers using a NFS share for the homedirs. The loadbalancer is using persistant connections (timeout 30min).
Upgrade...
Many, many bug fixes since the 1.0 release.
We made a upgrade to dovecot 1.0.0-1. It doesn't seem to help. We have the same problems whit this version. Any other ideas?
dovecot -n output again?
More than likely to do with NFS locking, since this is known to be finicky (I understand you can get it to work, but you have to get the locking set just right). I don't use NFS so won't be much more help than this - search the list (or google) on dovecot+NFS+lock...
good luck!
--
Best regards,
Charles
participants (2)
-
Charles Marcus
-
Christian Friedli