I have a CentOS 5.5 x86 server running on an OpenVZ VPS at 123systems.net.
I have postfix installed on this server and configured as a secondary MX.
I am frequently getting the following warning logged to maillog:
postfix/master[20377]: warning: process /usr/libexec/postfix/qmgr pid
20380 killed by signal 15
I've done some troubleshooting and can't figure out why qmgr is being
killed. It generally happens every 5 minutes, but not always. However
qmgr is set to wakeup at 300 second intervals in master.cf, so I think
what's happening there is that qmgr just gets run every 5 minutes no
matter what, and is simply getting killed shortly thereafter. It also
seems to get run whenever a connection is received on port 25, and again
gets killed shortly thereafter. So the 5 minute interval may be a red
herring. It may be better to say that qmgr just won't run for more than
a few seconds to a few minutes.
This is a secondary MX, so it's a pretty striaghtforward config, and I
have several other backup MX's which are running the *exact* same
config, except for the hostname being changed in main.cf, and they don't
suffer this problem. One is even another OpenVZ VPS at 123systems.net,
built from the same image!
To eliminate any possible config errors on my part, I rebuilt the VPS
from scratch from a clean Cent 5.5 image, ran "yum erase sendmail"
followed by "yum install postfix", and "service postfix start", made no
other configuration changes from the stock OS image, and even with that
completely clean stock config I still see the problem on this one VPS
(and only on this one VPS).
I would be inclined to think that this is just normal behavior (does
master kill qmgr if there is nothing for it to do?), but I'm puzzled as
to why I don't see it on any of my other (dozens) of postfix servers,
and why a warning is logged if this is normal behavior. I was of the
understanding that qmgr was supposed to keep running as a daemon even if
it has no work to do. Is this incorrect?
I've sent a few test messages through the server by telnetting directly
to port 25 and entering smtp commands, and they made it through to the
primary MX, so I *think* things are working, warning messages aside.
For now I've pulled the server out of my MX pool though.
I've Googled for this and can't find anyone else experiencing the same
thing, or even any mention anywhere to it being normal behavior. In
fact the only reference to this exact error I was able to find via
Google was a post I had to translate from Italian, which turned out to
be a gentleman with serer side antivrus software interfering with his
postfix setup. That explanation wasn't relevant to my setup though.
Nothing else in the logs seems to give any hint as to why this is happening.
Even though things seem to be working, seeing the error messages are
just driving me nuts in a very OCD kind of way. :-)
Any suggestions on how to further troubleshoot this?
- Jeff
