Re: [exim] v4.50 process stuck on select

Top Page
Delete this message
Reply to this message
Author: Michael Sprague
Date:  
To: Ian FREISLICH
CC: exim-users
Subject: Re: [exim] v4.50 process stuck on select
Ian FREISLICH wrote:
> Michael Sprague wrote:
>> Chris Lightfoot wrote:
>>> On Tue, Oct 24, 2006 at 12:41:16PM -0400, Michael Sprague wrote:
>>>     [...]
>>>> by now.  An strace doesn't show anything useful except that the process 
>>>> isn't doing much of anything:

>>>>
>>>> strace -f -p 29676
>>>> Process 29676 attached - interrupt to quit
>>>> select(10, [9], NULL, NULL, {2177, 130000}
>>>>
>>>> it looks like it's waiting for something, but how do I determine what?
>>>> This is becoming a big issue as I have to kill exim once in a while to
>>>> clean things up but it will eventually fill up again.
>>>>
>>>> Thanks for any pointers.
>>> use lsof or similar to identify -- in this case -- file
>>> descriptor #9 (presumably a socket or pipe), and see what
>>> clues that gives you. Connect to the process with gdb and
>>> get a backtrace, to see where it's getting stuck.
>> How could I forget about the power of lsof. :(
>>
>> lsof showed it was stuck waiting for spamd. I know where the issue is
>> now. Thanks for your help.
>
> Yes, version 4.50 has a timeout bug in the spam acl code. It will
> hang around for an hour IIRC. I don't think I ever found out what
> caused spamd to lock up.


Yes that was definitely the problem. I upgraded to 4.63 last night and
am watching things closely. So far so good but the heavy volume hasn't
really started to kick in yet. Now to get more resources for spamd. :)

thanks,
mikeS

-- 
Michael F. Sprague     | mfs@???
http://www.saneinc.net | Provider of SpamOnion anti-spam service
System and Network Engineering (SaNE), Inc