[Mailman-Users] Latency in 2.1.5

Jordan Hayes jmhayes at speakeasy.net
Mon Mar 14 17:54:21 CET 2005

I run a list with ~200 recipients that sees about 80-100 posts per day.
It's running 2.1.5 on Solaris 8 (pretty well patched) on a SunBlade 100
with plenty of memory and disk.  On a pretty regular basis, I see
messages take a long time to make it through the Mailman queue.  Today's
example was for 50 minutes.  I think.  It's hard to tell.  But here's
some sendmail logs:

Here's where it came in:

Mar 14 06:58:05 REMOVED sm-mta[22367]:
  [ID 801593 mail.info] j2EEvxPo022367:
  from=<REMOVED>, size=2487, class=0, nrcpts=1,
  msgid=<Pine.NEB.4.61.0503140943000.28291 at REMOVED>,
  proto=ESMTP, daemon=MTA, relay=REMOVED [REMOVED]

and here's the first batch it sent out:

Mar 14 07:48:30 REMOVED sm-mta[26377]:
  [ID 801593 mail.info] j2EFmTgK026377:
  from=REMOVED-bounces at REMOVED, size=3530, class=-30, nrcpts=24,
  msgid=<Pine.NEB.4.61.0503140943000.28291 at REMOVED>,
  proto=ESMTP, daemon=MTA, relay=localhost []

I looked at the logs and found this: the previous post finished at
07:48:29, suspiciously 1 second before the first batch of the next
message was sent.  This post finished at 08:08:09 ... 20 minutes later?
Could it really be taking 20 minutes to post?  The entry in the smtp log
says this:

Mar 14 08:08:09 2005 (25898) <Pine.NEB.4.61.0503140943000.28291 at REMOVED>
   smtp for 209 recips, completed in 1180.264 seconds

So, uh, yes I guess it did take that long.

I use weekly archives, and this week's archive directory only has about
a dozen messages in it.  The HTML archive was generated at 07:50:19 ...

So I guess there are two variables at work here that determine latency:
how long it takes to do one message, and how many messages are ahead of

Any ideas why this is taking so long or where else I could look to check
into it?



More information about the Mailman-Users mailing list