[Dovecot] How to remove duplicate GUID messages from storage?

[Daniel Parthey]

Timo Sirainen schrieb:
> On 21.4.2012, at 23.29, Daniel Parthey wrote:
> 
> > The following command
> > doveadm -v force-resync -u username at example.org INBOX
> > outputs lots of messages like this one, with different GUID
> > doveadm(username at example.org): Error: mdbox /mail/dovecot/example.org/username/mail/storage: Duplicate GUID 0b8a032d66a0924fb42c0000de5f8128 in m.55:45484041 and m.14:52173045
> 
> I think this is a force-resync bug and it shouldn't really complain about duplicates.
> Although I'm not entirely sure why with you it's complaining about them at all. I have anyway in TODO to look into this..

Maybe this is because I "manually" deleted all the mailboxes directories (rm -rf)
containing the duplicates, which one should better avoid with mdbox mailboxes... :(

So there is no "meta information" anymore, just the duplicate messages in the storage.
and dovecot would need to "guess" where these messages from the store belong.

> > How can I manually remove these identical, duplicate messages
> > from the storage to save space? Dovecot does not do it
> > automatically.
> 
> Perhaps force-resync + purge should do that, but currently it doesn't.

I already tried both, and purge is even running as a nightly cronjob.

> Also it would be nice if doveadm import didn't add duplicates in the first place.
> This is also something for which I have vague plans, because it would help dsync as well.

Thanks for looking into this.

I would really appreciate this idea of "duplicate GUID prevention" while
duplicate messages (with different GUIDs) should probably be allowed.
Maybe someone wants to store the same message several times...

Regards,
Daniel