On 19/06/2015 18:41, Mark wrote: > On 15/06/2015 18:04, Mark wrote: >> On 15/06/2015 17:46, Felix Zielcke wrote: >>> Am Sonntag, den 14.06.2015, 10:46 +0100 schrieb Mark: >>>> Since upgrading to Debian Wheezy >>> ... >>>> # 2.2.13: /etc/dovecot/dovecot.conf >>>> # OS: Linux 3.2.0-4-kirkwood armv5tel Debian 7.8 >>> >>> So it's dovecot from the backports. >>> Didn't you notice that since about 2 months there's now jessie out? >> >> I did. Based on my previous experience of upgrading Debian as soon as >> the new release was announced and having to roll everything back because >> of this problem I'm taking a more cautious approach this time. >> >>> That has 2.2.13 too but maybe it's an issue outside of dovecot in some >>> library. Or just a bug in the backport. >> >> Possibly. I'm not against doing the full upgrade to Jessie to see what >> effect that has. I've been running the backport for a while and the >> problem certainly hasn't got worse. I'll see if I can find some time in >> the next few days to try the upgrade. > > I haven't found the time to do the upgrade to Jessie yet but I have > tweaked my checking script to run more frequently. That gave me a > narrower window to look at in the logs and these messages appear at the > same time as the jump is introduced into the UID list: > > Jun 19 17:00:57 server02 dovecot: imap(mark): Error: Log synchronization > error at seq=2,offset=152 for > /secure/home/mark/Maildir/.Apache.Misc.committers/dovecot.index: Append > with UID 2427206830, but next_uid = 2685375011 > Jun 19 17:00:57 server02 dovecot: imap(mark): Warning: fscking index > file /secure/home/mark/Maildir/.Apache.Misc.committers/dovecot.index > > I've done some Googling but haven't (yet) found anything that might > match up. > > I'm hoping that the above might provide enough information to someone > more familiar with Dovecot than me to suggest what I might be able to do > to fix this. If not, I'll see what happens post Jessie upgrade.
One more information point. I still see the same symptoms after upgrading to Jessie. I haven't been able to see a pattern in the errors yet but I'll keep looking. Spending some time trying to isolate a test case is on my TODO list but until I find the time to do that any other suggestions welcome. Mark
