[Mailman-Users] Performance on mailman

[Peter Kofod]

 I checked out the article in 4.56.  Here is the output from my vmstat 1
20

procs -----------memory---------- ---swap-- -----io---- --system--
----cpu----
 r  b   swpd   free   buff  cache   si   so    bi    bo   in    cs us sy
id wa
 0  5 1910632  14280   1552  15020    7    4     2     6    3     2  0
1 65 34
 2  5 1909864  14716   1544  14708 3204    0  3204     0  523   389  0
1 50 49
 1  5 1908948  15160   1540  14992 3348    0  3348     0  541   453  0
3 50 47
 0  5 1907980  15612   1536  15056 3308    0  3308    60  533   421  0
0 48 51
 0  5 1907048  15848   1508  14716 3460    0  3460     0  572   445  0
2 50 49
 0  5 1905916  15652   1500  14768 3760    0  3760     0  568   446  0
2 50 49
 0  5 1905092  15672   1460  14620 3344    0  3344     0  571   423  0
1 50 49
 0  5 1904052  15716   1448  14268 3684    0  3684     0  575   442  0
1 50 49
 0  5 1903164  15780   1420  14252 3544    0  3544    20  580   449  0
1 50 49
 0  5 1902252  15804   1412  14100 3668    0  3668     0  563   434  0
2 50 48
 0  5 1901120  15712   1384  13968 3968    0  3968     0  575   499  0
1  3 95
 1  4 1900200  15852   1360  13620 3544    0  3544     0  543   417  0
1  0 99
 0  5 1899268  16228   1356  13568 3372    0  3372     0  544   429  0
1  0 99
 0  5 1898176  16376   1332  13304 3616    0  3616    20  538   440  0
2  0 98
 0  5 1897176  16416   1324  13176 3536    0  3536     0  551   440  0
2  0 98
 0  5 1896096  16600   1304  13104 3536    0  3536     0  547   424  0
1  0 100
 0  5 1895260  13388   1304  13212 3328    0  3328     0  533   427  0
2  0 98
 0  5 1894132  16460   1256  12756 4132    0  4132     0  600   462  0
2  0 99
 0  5 1893040  16188   1208  12776 3620    0  3620    20  582   477  0
2  0 98
 0  5 1892100  15132   1144  12612 3820    0  3820     0  554   456  0
1  0 99


My blocks in (bi) and swap in (si) seem very high compared to what the
FAQ says.  Furthermore, It looks like a lot of the processes are in a
wait state (far right), if I read this correctly.

Anyone have a clue what I did wrong?

Pete


-----Original Message-----
From: Mark Sapiro [mailto:msapiro at value.net] 
Sent: Saturday, October 28, 2006 8:05 PM
To: Peter Kofod; mailman-users at python.org
Subject: Re: [Mailman-Users] Performance on mailman

Peter Kofod wrote:

>I am doing something wrong.  I am running mailman with about 100 lists 
>(announce only).  A couple are pretty big (40K + addresses).  They, 
>however, are used infrequently.  The big list has one weekly mailing 
>and the others much less often.  Currently, there is NO mail queued or 
>being sent (tail -f on maillog is silent), yet doing a top on the 
>mailman process shows mailman consuming 1240M of virtual memory and a 
>task size
>(RES) of 811M.


See article 4.56 in the FAQ


>I am no Linux Performance Tuning expert, but the system seems pretty 
>pokey.  This is with no mail going out (and the occasional subscribe / 
>unsubscribe trickling in).
>
>When a mailing goes out, the system becomes virtually unresponsive, 
>usually resulting in a http 500 error.  If ssh in to the system, I am 
>eventually able to kill mailman, at which time the system perks right 
>up again.  System is a P4 with 1 GB of RAM.
>
>Mailman 2.19rc1


You will find some information by going to the FAQ wizard

>Mailman FAQ: http://www.python.org/cgi-bin/faqw-mm.py

and searching for performance. Article 4.56 will be one of the 16 hits.
Not all the others will be relevant, but there is good information
there.

I suspect (but it's only a suspicion) that during your 'unresponsive'
times, the thing that's going on is SMTP delivery of the message from
Mailman to the MTA. Tuning the MTA per the suggestions in the FAQ may
help. In particular, if the MTA is doing DNS verification of recipients
from Mailman, that's a real killer. There are suggestions in FAQ 4.11.

-- 
Mark Sapiro <msapiro at value.net>       The highway is for gamblers,
San Francisco Bay Area, California    better use your sense - B. Dylan