Berkeley DB4 can;t allocate memory errors

[Aidan Evans]

On Tue, 1 Oct 2002 at 13:44 Aidan Evans wrote to info-cyrus at andrew.cmu.ed=
u...

>  Anybody seen errors like this before and have suggestions about what t=
o do
>about them?
>
>	Sep 27 22:50:07 imap imapd[32227]: DBERROR db4: Lock table is out of
>	available locker entries
...
>	Oct 1 04:02:48 imap lmtpd[11979]: DBERROR: opening
>	/var/imap/deliver.db:  cyrusdb error
>
>They're happening thousands of times a day so I guess whatever the probl=
em
>is, it's not going to just go away.
>
>  I assume the databases are okay, just not being used (or are they brok=
en
>which is causing the error?).

  These errors went away after Cyrus was shut down and the system
rebooted.  The shutdown/reboot was not specifically intended to cure this=
,
it was for a disk reconfiguration.

  As part of this change all of /var/imap became /imap via "cp -ipr" and =
the
file system is ext3 (using the default journal=3Dordered) instead of ext2=
 (the
kernel is 2.4.9-34).  Now we have new errors.  First things like this
started:

	Oct 2 07:53:26 imap imapd[1471]: DBERROR: mystore: error storing
	X=C5=DC\230m)=EBx=A2=C6\177=A8Oo^Y\227+=DF&=AD: DB_PAGE_NOTFOUND: Reques=
ted page
	not found

	Oct 2 07:53:26 imap imapd[1471]: starttls: TLSv1 with cipher RC4-MD5
	(128/128 bits new) no authentication


Then, only once,

	Oct 2 08:13:00 imap lmtpd[5522]: duplicate_check:
	<200210021112.g92BCT119175 at Snoopy.UCIS.Dal.Ca> user.brownlee 0

	Oct  2 08:13:00 imap lmtpd[5522]: mystore: starting txn 2147706258

	Oct  2 08:13:00 imap lmtpd[5522]: DBERROR db4: PANIC: Invalid argument

	Oct  2 08:13:00 imap lmtpd[5522]: DBERROR: critical database situation

	Oct  2 08:13:00 imap master[1290]: process 5522 exited, status 75

followed by numerous

	Oct 2 08:13:02 imap lmtpd[5621]: DBERROR: error fetching
	<015a54c72e5b$7444c2b2$8ea06dd8 at gwadfo>: DB_RUNRECOVERY: Fatal
	error, run database recovery

	Oct 2 08:13:02 imap lmtpd[5621]: duplicate_check: error looking up
	<015a54c72e5b$7444c2b2$8ea06dd8 at gwadfo>/user.andrews: cyrusdb error

	Oct 2 08:13:02 imap lmtpd[5621]: duplicate_check:
	<015a54c72e5b$7444c2b2$8ea06dd8 at gwadfo> user.andrews 0

	Oct 2 08:13:02 imap lmtpd[5621]: DBERROR: mystore: error beginning
	txn:  DB_RUNRECOVERY: Fatal error, run database recovery

	Oct 2 08:13:02 imap lmtpd[5621]: duplicate_mark:
	<015a54c72e5b$7444c2b2$8ea06dd8 at gwadfo> user.andrews 1033557182

and

	Oct 2 08:13:17 imap lmtpd[5681]: DBERROR db4: region error
	detected; run recovery.

	Oct 2 08:13:17 imap lmtpd[5681]: DBERROR: dbenv->open '/imap/db'
	failed:  DB_RUNRECOVERY: Fatal error, run database recovery

	Oct 2 08:13:17 imap lmtpd[5681]: DBERROR: init /imap/db: cyrusdb
	error

	Oct 2 08:13:17 imap lmtpd[5681]: lmtpd: unable to init duplicate
	delivery database

  It feels as though deliver.db and tls_sessions.db have turned into
garbage.  What should I do?  Can I delete these files and will Cyrus then
re-create them cleanly so I can effectively start them over?  Can I do th=
is
while Cyrus is running?

name       : Cyrus IMAPD
version    : v2.1.9 2002/08/30 18:40:23
vendor     : Project Cyrus
support-url: http://asg.web.cmu.edu/cyrus
os         : Linux
os-version : 2.4.9-34smp-san
environment: Cyrus SASL 2.1.2
             Sleepycat Software: Berkeley DB 4.0.14: (November 18, 2001)
             OpenSSL 0.9.6b [engine] 9 Jul 2001
             CMU Sieve 2.2
             TCP Wrappers
             mmap =3D shared
             lock =3D fcntl
             nonblock =3D fcntl
             auth =3D unix
             idle =3D poll
             mboxlist.db =3D skiplist
             subs.db =3D flat
             seen.db =3D flat
             duplicate.db =3D db3-nosync
             tls.db =3D db3-nosync

Aidan Evans   | Networks & Systems
(902)494-3332 | University Computing & Information Services
              | Dalhousie University, Halifax, N.S., Canada