[Mailman-Users] python burning 100% cpu

[John Lange]

I think I have the problem solved. Acting on a hunch from what Jon Carns
said about the postfix configuration I completely stopped postfix and
restarted it.

Previously I had simply done a postfix reload.

Just for the benefit of anyone reading this thread in the future, I
believe the setting that made the difference was in postfix/main.cf. I
added "localhost" to the following line.

# RECEIVING MAIL
inet_interfaces = <MY IP>, localhost

Some time before I rebooted I had changed this from the default to
restrict postfix to listening to only a single IP instead of every IP on
the system (I had forgotten that I needed localhost).

The docs say "Note: you need to stop/start Postfix when this parameter
changes." In my haste I had missed that note and just did a reload
instead of a stop/start therefore postfix was still listening on
localhost and mailman worked fine until I rebooted.

On the flip side I still did not see the note when I added localhost
back in and again failed to do a stop/start.... *sigh*

Clearly my fault for not reading the docs but if any developers are
reading this; my recommendation would be that mailman should report a
failed connection attempt to localhost as a serious problem and log it
in syslog, or at the very minimum it's own logs. quietly burning 100%
CPU should not be acceptable.

The only reason I even noticed something was wrong at all was because I
run a remote gkrellm client on my desktop for the server. Without that
it probably would have been burning away for days or weeks before I
happened to check top or something.

It is also not clear to me why this would cause 100% CPU burn on startup
since I wasn't sending anything to any lists at the time. Perhaps
something was still queued ?

Thanks to those who responded.

John Lange

On Thu, 2003-09-11 at 14:27, Jon Carnes wrote:
> I don't think those errors mean too much (unless you upgraded and it
> hasn't finished the upgrade...)
> 
> If your MTA is Postfix, be sure that you have followed all the advise in
> the README.POSTFIX file.  Specifically make sure that local bounces are
> not set to a 4xx error code (which means retry and retry and retry and
> retry.....you get the idea).
> 
> Other than that, you might want to turn off archiving for the nonce and
> see if that helps. Some slower servers with IDE drives have a very hard
> time with this - especially if they don't have enough RAM.
> 
> Those are the two biggest causes of CPU over-run.
> 
> Best of Luck - Jon Carnes
> 
> On Thu, 2003-09-11 at 13:27, John Lange wrote:
> > I'm new to Mailman and I'm having a major problem getting it to start.
> > 
> > I had it setup and working and then I needed to reboot the server for a
> > kernel upgrade.
> > 
> > Now when I start mailman it causes python to burn 100% CPU until mailman
> > is stopped.
> > 
> > logs show nothing out of the ordinary.
> > 
> > logs/qrunner
> > Sep 11 12:20:40 2003 (4450) ArchRunner qrunner started.
> > Sep 11 12:20:40 2003 (4452) CommandRunner qrunner started.
> > Sep 11 12:20:40 2003 (4453) IncomingRunner qrunner started.
> > Sep 11 12:20:40 2003 (4451) BounceRunner qrunner started.
> > Sep 11 12:20:40 2003 (4455) OutgoingRunner qrunner started.
> > Sep 11 12:20:40 2003 (4456) VirginRunner qrunner started.
> > Sep 11 12:20:40 2003 (4454) NewsRunner qrunner started.
> > 
> > Nothing in syslog either.
> > 
> > I ran check_perms and everything checks out fine.
> > 
> > check_db -va though shows some errors:
> > 
> > List: announce
> >    /usr/local/mailman/lists/announce/config.pck: okay
> >    /usr/local/mailman/lists/announce/config.pck.last: okay
> >     [Errno 2] No such file or directory:
> > '/usr/local/mailman/lists/announce/config.db'
> >     [Errno 2] No such file or directory:
> > '/usr/local/mailman/lists/announce/config.db.last'
> > List: mailman
> >    /usr/local/mailman/lists/mailman/config.pck: okay
> >    /usr/local/mailman/lists/mailman/config.pck.last: okay
> >     [Errno 2] No such file or directory:
> > '/usr/local/mailman/lists/mailman/config.db'
> >     [Errno 2] No such file or directory:
> > '/usr/local/mailman/lists/mailman/config.db.last'
> > 
> > Could this be the problem? Are those files critical? If so, what
> > happened to those files? How can they be re-created?
> > 
> > I've hit a brick wall here. With nothing being logged I don't know where
> > to go from here.
> > 
> > Thanks for any help.