huge mdbox broken / Log synchronization error

[Aki Tuomi]

> On 06/05/2021 14:22 Florian Lohoff <f at zz.de> wrote:
> 
>  
> On Thu, May 06, 2021 at 02:11:16PM +0300, Aki Tuomi wrote:
> > > doveadm(inbox at wasteland.rfc822.org): Warning: Transaction log file /var/vmail/wasteland.rfc822.org/inbox/mdbox/storage/dovecot.map.index.log was locked for 15023 seconds (rotating while syncing)
> > > doveadm(inbox at wasteland.rfc822.org): Error: Corrupted transaction log file /var/vmail/wasteland.rfc822.org/inbox/mdbox/dovecot.list.index.log seq 4: indexid changed: 1620283336 -> 1620289271 (sync_offset=0)
> > > doveadm(inbox at wasteland.rfc822.org): Warning: fscking index file /var/vmail/wasteland.rfc822.org/inbox/mdbox/storage/dovecot.map.index
> > > 
> > > I have no clue if this is a:
> > > 
> > > "I have repaired everything - be happy"
> > > 
> > > or
> > > 
> > > "Unfixable broken - Better use your backup"
> > > 
> > > 
> > > So i guess i remove all dovecot.map.index* files now and restart over.
> > 
> > I would suggest trying to fix it by adding nologin attribute to the account to avoid concurrent access before doing those things.
> 
> That does help me interpret the force-resync output?!?!
> 
> I stopped new delivery yesterday noon already by putting all mails on
> hold in postfix (100K+ now already).
> 
> I made imap login now impossible by changing the password.
> 
> Flo

Sorry, I ment that it really looks like, based on the logs, that something else was hoarding your indexes for 15023 seconds, and I can't really think anything else than some imap process being there. Or you are running it twice.

Aki