On Thu, 06 May 2010 00:26:36 -0500, Noel Jones <njo...@megan.vbhcs.org> wrote: > On 5/5/2010 10:00 PM, John Robinson wrote: >> Hello all, >> >> As this is my first post to this list, please bear with me if my >> description or supplied info is not all that is required. >> >> I have 2 mail gateways running on Debian Lenny. I have installed and >> configured the following according to this guide : >> http://www200.pair.com/mecham/spam/spamfilter20090215.html#notes >> >> Debian - 5.0.4 (kernel 2.6.26-2-amd64) >> Postfix - 2.5.5-1.1 >> amavis - 2.6.4 >> clamav - 0.95.3+dfsg-1 >> spamassassin - 3.2.5-2+Lenny2 >> postgrey - 1.31-3.2 >> >> My problem is that about once a week the servers (at different times and >> not consistently) will start queuing mail >> and will not route it out. They still accept inbound smtp connections to >> port 25 but will not pass them on to amavis ?!? >> >> In the log files I find entries such as the one below : >> >> May 5 06:46:27 mailgateway00 postfix/smtp[1525]: DC2C114C68B: >> to=<u...@domain.com>, relay=127.0.0.1[127.0.0.1]:10024, delay=300, >> delays=0.13/0.01/300/0, dsn=4.4.2, status=deferred (conversation with >> 127.0.0.1[127.0.0.1] timed out while receiving the initial server >> greeting) >> >> executing "mailq" on the server reveals a similar message in the mail >> queue >> >> (delivery temporarily suspended: conversation with 127.0.0.1[127.0.0.1] >> timed out while receiving the initial server greeting) >> u...@domain.com >> >> The only way I have been able to fix the problem so far is to restart the >> whole server. Mail will then route until the next time this happens. > > Just to clarify, I'm assuming you're referring to amavisd-new > and not some other variant of amavis, because all the others > are dead projects and should be avoided. > > (continue, assuming all references are for amavisd-new) > > General debug strategy: > - do you have plenty of RAM? Spamassassin and clam can really > chew up some megs. > - any errors in the log from about the time mail stops? Look > before the "deferred" messages start showing up. > - what's the last thing amavis logs? > - can you telnet to the amavis port? > - does the amavisd-nanny program tell you anything > interesting? (run it a few times when things are normal so > you know what it should look like). > - does restarting amavisd get mail flowing within a few > minutes? May require a "postfix flush" or waiting several > minutes for postfix to realize that the destination is no > longer dead. Note: frequent "postfix flush" with a full queue > is very bad for performance, so use sparingly. > > I find it useful to run one "extra" amavis server process so > that there will always be one free for testing. > > If amavisd appears unresponsive, increasing the amavisd log > level may help you pinpoint where the trouble is. > > > HTH. > > -- Noel Jones > > ------------------------------------------------------------------------------ > _______________________________________________ > AMaViS-user mailing list > AMaViS-user@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/amavis-user > AMaViS-FAQ:http://www.amavis.org/amavis-faq.php3 > AMaViS-HowTos:http://www.amavis.org/howto/
Thanks for the prompt response Noel. You are correct, I am referring to amavis-new. The systems I am running have 4Gb of RAM and usage seems to sit at around 1.5gig, I have never seen any swap usage. I have had a look at the log files on both servers just before the error started occurring and the only thing slightly out of the ordinary I can see is an entry like the below : timeout after END-OF-MESSAGE from localhost[127.0.0.1] Searching online for this indicates that it is to do with connection caching and likely not relevant to this issue. Is this correct ? The last thing amavis seems to log looks like normal message processing. Let me know if you would like to see a log snippet of around the time the error starts. I can telnet to the amavis port (10025) when the error is occurring. I have now got amavisd-nanny running and will run it again when the error reoccurs to see if it reveals anything. Restarting amavis does not seem to alleviate the problem. Only restarting the whole server seems to work for me at the moment. Also I have now increase the number of amavis processes that run by one. I'll let you know how I got at the next reoccurrence of this error. Thanks for your help. Regards John ------------------------------------------------------------------------------ _______________________________________________ AMaViS-user mailing list AMaViS-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/amavis-user AMaViS-FAQ:http://www.amavis.org/amavis-faq.php3 AMaViS-HowTos:http://www.amavis.org/howto/
