Skip to Content.
Sympa Menu

en - Re: [sympa-users] anomalous Sympa distribution behavior

Subject: The mailing list for listmasters using Sympa

List archive

Chronological Thread  
  • From: David Verdin <address@concealed>
  • To: Thomas Berry <address@concealed>
  • Cc: "address@concealed" <address@concealed>
  • Subject: Re: [sympa-users] anomalous Sympa distribution behavior
  • Date: Thu, 17 Sep 2009 14:43:59 +0200

Hi Thomas,

Was Sympa handling a single message or several? I wondered if you could not have suffered a kind of attack : somebody send thousands of messages to the sympa robot, so that it spends a lot of time trying to execute commands instead of sending messages. This could explain why it managed to send a message from time to time.

By the way, if it's nothinf critical for you, I'm interrested in reading the script you used to analyze your logs. That could be very interesting for us, to narrow on what is the most time consuming in Sympa.

Regards,

David

Thomas Berry a écrit :
Sympa version 5.3
OS: RedHat Release 4.4

Has anyone experienced this behavior with Sympa:

We had four occurrences over a 9 hour period where the Sympa command process would mostly stop processing messages from the incoming spool (msg). I say "mostly" because during this outages a small percentage of messages to mailing lists continued to be processed by Sympa (command) and distributed without a delay. During the outages, Sympa was restarted, which appeared to have no effect on Sympa's recovery.


I wrote a maillog and message log trolling script to link the delivery path of each of the distributed messages to show the distribution timeline. Here is an example of the outage I see (the time durations are in seconds):

ibs-help
sendmail recd Sep 14 07:23:23
sendmail queued Sep 14 07:23:24 delay: 1
sympa(command) recd Sep 14 07:44:12 delay: 1248
sympa(message) recd Sep 14 07:44:16 delay: 4
sympa(message) dist Sep 14 07:44:16 delay: 0
sendmail dist Sep 14 07:44:18 delay: 2
duration: 1255

As you can see in this example, the time it took the sympa(command) process to retrieve the message from the msg spool was 1248 seconds. That was when the "outage" itself resolved and the messages "stuck" in the queue were quickly distributed.

I have never seen this happen before, nor has it occurred after the fourth instance.

There have been no recent changes to this system's hardware or software.

Implementation notes:
MySQL database, LDAP server, and mounted filer used for the MhonArc repository all reside external to the Sympa server. Inbound and Outbound messages to and from Sympa come from a cluster of Sendmail, Inc MTA appliances.


--
David Verdin
Comité réseau des universités




Archive powered by MHonArc 2.6.19+.

Top of Page