[Mailman-Users] database error message when upgraded to 2.1.2
Richard Barrett
r.barrett at openinfo.demon.co.uk
Fri May 2 08:23:40 CEST 2003
At 04:41 02/05/2003, NOW Website Coordinator wrote:
>At 09:21 PM 05/01/2003 -0400, Jon Carnes wrote:
>>On Thu, 2003-05-01 at 18:19, NOW Website Coordinator wrote:
>>
>> > For some reason, both on our old Solaris site and now on our new Linux
>> > site, the Outgoing Runner takes up 70- 90% of our CPU. I even told it to
>> > be "nice" on Linux.
>> >
>>I'm not seeing that on any of my sites. Mailman 2.0.13 and Mailman
>>2.1.1 all work very nicely and only eat up CPU or RAM while crunching
>>away on messages. Once the queues are empty Mailman drops out of site
>>on my "top" command.
>>
>>Take a look in some of the directories:
>> ls -l ~mailman/data
>> ls -l ~mailman/qfiles/*
>
>
>There are only 6 files in any qfile folders at the moment and only a few
>in data (all of our lists are newsletter type lists).
>
>I didn't even do the monthly reminder for the last few months, because
>Mailman was just too much for our machine. Which is why we switched to a
>machine 3x faster, but it hasn't helped.
>
>
>>You can run each of the qrunners separately (one at a time) and see
>>which one eats up the most cpu. My guess is the archiver.
>
>We use Mhonarc, and the job that is taking up all the CPU is the outgoing
>runner:
So, presumably you have configure Mailman to deliver archivable message by
command directly to MHonarc rather than having MHonarc get the material via
list subscriptions.
It is a pity you did not mention the use of MHonarc to start with. The
error message you quotedin your earlier post as being generated on
mailmanctl start is eerily similar to one described in a file I found in
MHonarc's distribution MHonArc2.5.13/doc/app-diagnos.html#warnings; I do
not use MHonarc myself but grepping through an expanded distribution found it.
<quote>
Warning: Database (dbversion) != program (prgversion) version
Indicates that the version of MHonArc updating an archive is different from
the version of MHonArc that created the database file. Problems can arise
if the database file changes in format from different version of MHonArc.
See the release notes of the MHonArc distribution if changes in the databse
format will effect older archives.
</quote>
It seems a reasonable speculation that you have updated MHonarc at some
point and this is the cause of the problem and possibly this problem has
nothing to do with Mailman but that MHonarc is grinding exceeding small? It
also explains why the problem appeared to followe Mailman. But maybe it is
following MHonarc which is attached to Mailman??
>21689 ? RN 541:55 qrunner /usr/local/mailman/bin/qrunner
>--runner=OutgoingRunner:0:1 -s
>
>Although it's not doing any message sending other than occasional
>sub/unsub messages.
>
>The other mailman jobs are not taking up that much CPU, although several
>are using over 50 megs of RAM (and not doing anything, as I said). From
>top on the Linux machine:
>
> PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
>21689 mailman 16 4 57292 55M 3896 R N 88.0 5.5 547:54 python
>21683 mailman 11 4 4672 1904 1896 S N 0.0 0.1 0:00 mailmanctl
>21684 mailman 11 4 14272 13M 1876 S N 0.0 1.3 0:06 python
>21685 mailman 11 4 46336 39M 3352 S N 0.0 3.9 4:24 python
>21686 mailman 11 4 51760 50M 2020 S N 0.0 5.0 2:26 python
>21687 mailman 11 4 20504 19M 8944 S N 0.0 1.9 1:11 python
>21688 mailman 11 4 4652 2768 1812 S N 0.0 0.2 0:05 python
>21690 mailman 11 4 51564 50M 1984 S N 0.0 4.9 6:39 python
>
>And over on the Solaris machine which isn't doing anything at all (nothing
>incoming, no qfiles):
>
>
> PID USERNAME THR PRI NICE SIZE RES STATE TIME CPU COMMAND
> 20332 mailman 4 30 0 73M 72M run 967:30 72.16% python
> 20333 mailman 4 59 0 47M 46M sleep 2:52 0.00% python
> 20328 mailman 4 58 0 16M 15M sleep 0:07 0.00% python
> 20331 mailman 4 59 0 7472K 6256K sleep 0:01 0.00% python
> 20327 mailman 4 59 0 7432K 6216K sleep 0:01 0.00% python
> 20330 mailman 4 59 0 7432K 6216K sleep 0:01 0.00% python
> 20329 mailman 4 59 0 7432K 6216K sleep 0:01 0.00% python
> 20326 mailman 3 58 0 7528K 3048K sleep 0:00 0.00% mailmanctl
>
>(I just asked top to display mailman jobs, but it's still the top job on
>both machines even though the Linux one is also running a catalog, four
>web servers, email, etc.).
>
>
>>What version of python are you using?
>
>Python 2.2.2
>
>We now have 1 gigabyte of RAM, and 1.26GHz CPU.
>
>Thanks for helping trying to figure this out. I really don't understand
>why it's running and running but would love to get it under control.
More information about the Mailman-Users
mailing list