[SOLVED] Amavisd hangs

[vginders]

When enabling AV/AS, on a certain server (5.0, upgraded to last version today) I have a runnig problem wih AMAVISD: amavisd runs on 100% processor, and seems to hang continuously. All mail is getting deferred as it cannot be passed from postfix to amavis. Disabling AV/AS is a workaround, but as this issue remains in 5.0, I'd like to find a solution.

I have been searching different fora, bugzilla's, wiki's, but I'm afraid I can't seem to find any solution on why this is happening. I already enabled some verbose loggings (smtp -v from postfix to amavis, log_level 5 in amavis), but this does not give any extra info on why Amavis itself is running loose.

Jan  4 17:00:05 deepthroat postfix/qmgr[6005]: 5EC4F584E1: to=, relay=none, delay=0.14, delays=0.12/0.02/0/0, dsn=4.3.0, status=deferred (mail transport unavailable)

Any pointers on how to start troubleshoting this?

[quanah]

I'd start by verifying that the amavis user can bind to the LDAP server.

[vginders]

Originally Posted by quanah

I'd start by verifying that the amavis user can bind to the LDAP server.

Thanks. Any pointer on the syntax? I can't find any info on this.

[quanah]

zmlocalconfig -s ldap_amavis_password

to get the password. Then:

ldapwhoami -x -h ldaphost.FQDN -ZZZ -D "uid=zmamavis,cn=appaccts,cn=zimbra" -W

It'll prompt for the password. If the result is success, then the password is correct in the LDAP server DB. If you get an error, please post it.

[vginders]

No problem here:

zimbra@deepthroat:~$ ldapwhoami -x -h deepthroat.cafe.denbeirensystems.be -ZZZ -D "uid=zmamavis,cn=appaccts,cn=zimbra" -W
Enter LDAP Password: 
dn:uid=zmamavis,cn=appaccts,cn=zimbra
Result: Success (0)
zimbra@deepthroat:~$

[quanah]

Okay, so that excludes two issues -- A password problem, or a TLS problem (The -ZZZ option forces the use of TLS). I'm not sure what then would cause Amavis to hang, I'll see if others have some idea.

--Quanah

[vginders]

I might add that this behaviour started when upgrading from the last beta to the RC, and survived when upgrading to the definitive december FOSS version.
I have a similar other server, same OS, same kind of config, which does not have the problem. It somehow started on this one (customer) setup

[quanah]

Hm, have you tried clearing the amavis DB on the problematic server? (See /opt/zimbra/amavisd-new-2.5.2/db), and verified that /opt/zimbra/amavisd is linked to the right release?

IIRC one of the RCs didn't migrate amavis correctly.

[vginders]

Soft link was ok, after moving the logs en re-enabling amavis I get this difference in db files:

zimbra@deepthroat:~/amavisd-new-2.5.2$ ls -l db*
db:
total 408
-rw-r----- 1 zimbra zimbra  12288 2008-01-08 12:21 cache.db
-rw-r----- 1 zimbra zimbra   4096 2008-01-08 12:21 cache-expiry.db
-rw-r----- 1 zimbra zimbra  16384 2008-01-08 12:21 __db.001
-rw-r----- 1 zimbra zimbra 278528 2008-01-08 12:21 __db.002
-rw-r----- 1 zimbra zimbra 450560 2008-01-08 12:21 __db.003
-rw-r----- 1 zimbra zimbra  12288 2008-01-08 12:21 nanny.db
-rw-r----- 1 zimbra zimbra  12288 2008-01-08 12:21 snmp.db

db.old:
total 380
-rw-r----- 1 zimbra zimbra  12288 2008-01-04 16:17 cache.db
-rw-r----- 1 zimbra zimbra   4096 2008-01-04 16:17 cache-expiry.db
-rw-r----- 1 zimbra zimbra  16384 2008-01-04 16:17 __db.001
-rw-r----- 1 zimbra zimbra 278528 2008-01-04 16:17 __db.002
-rw-r----- 1 zimbra zimbra 450560 2008-01-04 16:17 __db.003
-rw-r----- 1 zimbra zimbra  12288 2007-11-29 18:01 __db.snmp.db

I looks solved now. I monitor this a day and give feedback later.

Thanks!