Hi!

We have some strange replication issues. Some mails are duplicated when expunged. This bug is quite hard to reproduce, it just happens occasionally. It seems somehow related to sieve. When i deleted the sieve script i could not reproduce this behavior.

Here are some logs, mjm-100010@mydomain.com is an alias for mjm@mydomain.com:

# Master Jun 23 00:51:41 dobby4 dovecot: lmtp(mjm-100010@mydomain.com): copy from <lmtp DATA>: box=INBOX, uid=7816, msgid=20160623005132EE517444BB$3A04DF25DD@ABRAHAM, size=8933 Jun 23 00:51:41 dobby4 dovecot: lmtp(mjm-100010@mydomain.com): 4N23HaMVa1fpmQEAJEBNLA: sieve: msgid=20160623005132EE517444BB$3A04DF25DD@ABRAHAM: stored mail into mailbox 'INBOX' Jun 23 04:52:57 dobby4 dovecot: pop3(mjm@mydomain.com): expunge: box=INBOX, uid=7816, msgid=20160623005132EE517444BB$3A04DF25DD@ABRAHAM, size=8933 Jun 23 04:52:57 dobby4 dovecot: dsync-local(mjm@mydomain.com): save: box=INBOX, uid=7817, msgid=20160623005132EE517444BB$3A04DF25DD@ABRAHAM, size=8933 Jun 23 04:55:21 dobby4 dovecot: pop3(mjm@mydomain.com): expunge: box=INBOX, uid=7817, msgid=20160623005132EE517444BB$3A04DF25DD@ABRAHAM, size=8933

# Replica Jun 23 00:51:42 dobby5 dovecot: dsync-server(mjm-100010@mydomain.com): save: box=INBOX, uid=7816, msgid=20160623005132EE517444BB$3A04DF25DD@ABRAHAM, size=8933 Jun 23 04:52:57 dobby5 dovecot: dsync-server(mjm@mydomain.com): copy from INBOX: box=INBOX, uid=7817, msgid=20160623005132EE517444BB$3A04DF25DD@ABRAHAM, size=8933 Jun 23 04:52:57 dobby5 dovecot: dsync-server(mjm@mydomain.com): expunge: box=INBOX, uid=7816, msgid=20160623005132EE517444BB$3A04DF25DD@ABRAHAM, size=8933 Jun 23 04:55:22 dobby5 dovecot: dsync-server(mjm@mydomain.com): expunge: box=INBOX, uid=7817, msgid=20160623005132EE517444BB$3A04DF25DD@ABRAHAM, size=8933

The problem is the "dsync-server(mjm@mydomain.com): copy from INBOX: box=INBOX, uid=7817" on dobby5 at 04:52:57. This creates a new uid 7817 and the new mail is replicated back to dobby4. But i have no idea what triggers this copy.

Any ideas?

The dovecot version is 2.2.24.

Thanks Dennis