FATAL: lmtpd: unable to init duplicate delivery database

[Dan White]

On 07/04/10 00:45 +0200, Marcus wrote:
>I've restarted cyrus and got a "recovery /var/lib/cyrus/deliver.db:
>found partial txn, not replaying". Delivering is working fine and
>dupelim check too. Could someone confirm, that Simon's idea to clean
>up /var/lib/cyrus/db and remove  deliver.db is _not_ dangerous or do I
>have to run some kind of reconstruct after that?
>
>-------------------------
>Apr  7 00:18:37 dexter cyrus/master[5510]: process started
>Apr  7 00:18:37 dexter cyrus/master[5511]: about to
>exec /usr/sbin/ctl_cyrusdb
>Apr  7 00:18:38 dexter cyrus/ctl_cyrusdb[5511]: recovering cyrus
>databases
>Apr  7 00:18:38 dexter cyrus/ctl_cyrusdb[5511]: skiplist:
>recovered /var/lib/cyrus/mailboxes.db (1289 records, 165424 bytes) in 0
>seconds
>Apr  7 00:18:38 dexter cyrus/ctl_cyrusdb[5511]: skiplist:
>recovered /var/lib/cyrus/annotations.db (0 records, 144 bytes) in 0
>seconds
>Apr  7 00:18:38 dexter cyrus/ctl_cyrusdb[5511]: done recovering cyrus
>databases
>Apr  7 00:18:38 dexter cyrus/master[5512]: about to
>exec /usr/sbin/cyr_expire
>Apr  7 00:18:38 dexter cyrus/cyr_expire[5512]: skiplist
>recovery /var/lib/cyrus/deliver.db: found partial txn, not replaying
>Apr  7 00:18:39 dexter cyrus/cyr_expire[5512]: skiplist:
>recovered /var/lib/cyrus/deliver.db (12426 records, 2651376 bytes) in 1
>second
>Apr  7 00:18:39 dexter cyrus/cyr_expire[5512]: duplicate_prune: pruning
>back 3 days
>-------------------------

You should probably checkpoint your databases just before shutting down to
be safe:

/usr/sbin/ctl_cyrusdb -c

as the cyrus user

-- 
Dan White