db3 database error

Ramji Venkateswaran rv at uiop.org
Sat Dec 6 08:51:06 EST 2003


I hope you're able to shed some light on this particular problem I'm
seeing. Regular as clockwork, every day, I see:

Dec  6 03:00:15 imaphost cyrus/master[19337]: set maximum file descriptors to 256/256
Dec  6 03:00:15 imaphost cyrus/master[19337]: about to exec /usr/sbin/ctl_cyrusdb
Dec  6 03:00:15 imaphost cyrus/ctl_cyrusdb[19337]: checkpointing cyrus databases
Dec  6 03:00:15 imaphost cyrus/ctl_cyrusdb[19337]: DBERROR: archive /var/lib/cyrus/db: cyrusdb error
Dec  6 03:00:15 imaphost cyrus/ctl_cyrusdb[19337]: archiving log file: /var/lib/cyrus/db/log.0000000001
Dec  6 03:00:15 imaphost cyrus/ctl_cyrusdb[19337]: error opening /var/lib/cyrus/db.backup1//log.0000000001 for writing (33184)
Dec  6 03:00:15 imaphost cyrus/ctl_cyrusdb[19337]: DBERROR: error archiving log file: /var/lib/cyrus/db/log.0000000001
Dec  6 03:00:15 imaphost cyrus/ctl_cyrusdb[19337]: DBERROR: archive /var/lib/cyrus/db: cyrusdb error
Dec  6 03:00:15 imaphost cyrus/ctl_cyrusdb[19337]: done checkpointing cyrus databases
Dec  6 03:00:15 imaphost cyrus/master[4363]: process 19337 exited, status 0

This then leads onto such fun as all mail failing to be delivered,
exim/lmtpd blocking on a read and the messages as below:

Dec  6 06:30:15 imaphost cyrus/master[25323]: set maximum file descriptors to 256/256
Dec  6 06:30:15 imaphost cyrus/master[25323]: about to exec /usr/sbin/ctl_cyrusdb
Dec  6 06:30:15 imaphost cyrus/ctl_cyrusdb[25323]: checkpointing cyrus databases
Dec  6 06:30:15 imaphost cyrus/ctl_cyrusdb[25323]: DBERROR: archive /var/lib/cyrus/db: cyrusdb error
Dec  6 06:30:15 imaphost cyrus/ctl_cyrusdb[25323]: DBERROR db3: region error detected; run recovery.
Dec  6 06:30:15 imaphost cyrus/ctl_cyrusdb[25323]: DBERROR: dbenv->open '/var/lib/cyrus/db' failed: DB_RUNRECOVERY: Fatal error, run database recovery
Dec  6 06:30:15 imaphost cyrus/ctl_cyrusdb[25323]: DBERROR: init /var/lib/cyrus/db: cyrusdb error
Dec  6 06:30:15 imaphost cyrus/ctl_cyrusdb[25323]: done checkpointing cyrus databases
Dec  6 06:30:15 imaphost cyrus/master[4363]: process 25323 exited, status 1

The only appropriate thing to do to resolve the problem at a superficial
level - that I've found, hence my email - is:

	/etc/init.d/cyrus21 stop
	/usr/bin/db3_recover  -v -h /var/lib/cyrus/db/
	/etc/init.d/cyrus21 start

And I'm happy to do this every hour if needs be, *but* it does break
people who use mail clients that utilise persistent connections (i.e.
the better ones) - so I'm not eager to simply shove this in my crontab
and cope.

I'm hoping there's an obvious solution that my googling has missed, as
this does seem to happen with regularity.

I've put various configuration data at 
	http://www.uiop.org/~rv/info-cyrus.help.html

Hope you can help,

Ramji


-- 
Ramji Venkateswaran
My views are my own, not necessarily those of the people 
that I am affiliated with or work for.

If you tell the truth you don't have to remember anything.
		-- Mark Twain




More information about the Info-cyrus mailing list