4

We've got mail accumulating in our Leopard Server's queue and not sure exactly why. This machine has required little maintenance over the years so I'm hoping someone here spot the obvious and save us some time. Let me know what other information would be helfull. Server appears to be functioning normally except for "clogged" queue and the following error associated with each "trapped" message:

Looking at messages in the queue each one states something like this: 

Message ID: 4213C3B8B3F
Date:       October 27, 2009 11:33:27 AM
Size:       1824
Sender:     user@gmail.com
Recipient(s) & Status:
----------------------
user@ourdomain.com:
    connect to 127.0.0.1[127.0.0.1]: Connection refused

Under Settings>Relay we have checked Accept SMTP relays only from these hosts and networks:

127.0.0.0/8
10.0.1.0/24

The mail in queue is addressed to users whose accounts are on this server. Mail.app on the client appears to be functioning normally and checking checking mail on the server. We did add a virtual domain some time ago but all that was working fine for some time... This just started happening recently...any ideas?

Edit: toggling the filter services on and off seems to have fixed this except for 2 remaining queued messages that show "mail transport unavailable" as an error!?!

Meltemi
  • 579
  • 2
  • 11
  • 24

2 Answers2

1

We see this on occasion on our Mac OS X 10.5 Leopard Server mail server as well, usually after restarting the Mail service (I actually attribute it to a mailbfr issue, but haven't fully traced it to the root cause). What happens is that /var/amavis/tmp goes missing, causing amavisd to fail, thus preventing delivery. Simply recreating the tmp directory and restarting the Mail service works for us:

sudo mkdir /var/amavis/tmp
sudo chown _amavisd:_amavisd /var/amavis/tmp

As far as cleaning out the queue, I'd suggest getting comfortable with the postqueue & postsuper commands. postqueue -p will print the queue. sudo postsuper -r ALL will re-queue all the messages for delivery. I'd suggest not running the latter when you have a massive number of messages in the queue, but once the queue clears out and you just have a couple that haven't been resent yet.

morgant
  • 1,470
  • 6
  • 23
  • 33
  • Your observation that /var/amavis/tmp goes missing from time to time might be related to your server using an SSD and you having disabled access times to be written. Then amavisd thinks it didn't actually need the temp directory and removes it. (Yes this behaviour is extremely dumb.) – MacLemon Sep 06 '10 at 11:11
  • Bingo! We're definitely running on SSDs w/access times disabled, so that is _very_ good to know. – morgant Sep 07 '10 at 20:59
0

Could it have something to do with content filtering? http://discussions.apple.com/thread.jspa?messageID=6847070

Bart Silverstrim
  • 31,172
  • 9
  • 67
  • 87
  • I found that link on Apple's site before the OP. It's not resolved so I'm a bit wary of trying what the last poster recommends. As for the first response - I'm not sure what he means by toggling the "content filter" in Server Admin? Under filters both Junk & Virus are turned OFF. There is no "content" filter that I can find?!? – Meltemi Oct 27 '09 at 21:05
  • I ended up toggling both the Junk & Virus filters on, save, off again, save. and now things appear to be working and MOST of the queued mail has been delivered. However, one user still has inbound mail stuck in the queue, "mail transport unavailable", but is receiving NEW messages normally?!? – Meltemi Oct 27 '09 at 22:53
  • It keeps coming back to something with Amavisd or spam processing when I look it up for postfix (http://www.linuxquestions.org/questions/linux-server-73/mail-transport-unavailable-error-753514/). But altering conf files directly can break things in OS X. This (http://discussions.apple.com/thread.jspa?threadID=2165968&tstart=30) also seems to support that Amavis is having issues. – Bart Silverstrim Oct 28 '09 at 11:34