On Tuesday 15 July 2003 13:40, Jesse Guardiani wrote:
> On Tuesday 15 July 2003 13:04, Tom Walsh wrote:
> > I believe that I am seeing a correlation between these
> > events and a log entry of "pid 37869 (sqwebmail), uid 90: exited on
> > signal 10"
>
> Don't count on it. I get those occassionally with FreeBSD 4.8 and the
> latest production build of sqwebmail, but my customers don't have any
> problems accessing sqwebmail from IE 5.5.
Here is a log snippet from when I was running 1.5.1:
Jul 10 09:24:46 chortos /kernel: pid 75543 (sqwebmail), uid 89: exited on signal 10
Jul 10 09:31:27 chortos /kernel: pid 77768 (sqwebmail), uid 89: exited on signal 10
Jul 10 15:53:55 chortos /kernel: pid 16391 (sqwebmail), uid 89: exited on signal 11
Jul 10 16:06:05 chortos /kernel: pid 20584 (sqwebmail), uid 89: exited on signal 11
With sqwebmail 1.5.1, these dying processes would eat 100% CPU for a period
of time (30 min? An hour? 5 hours?) before dying. And usually there are two or
three at a time, then none for a few days.
I upgraded to the 20030629 development release on Jul 10 at around 17:00 hours,
and so far I have only this in my log with the new version:
Jul 13 20:52:08 chortos /kernel: pid 94327 (sqwebmail), uid 89: exited on signal 11
And I noticed this in the ChangeLog:
2003-04-03 Mr. Sam <[EMAIL PROTECTED]>
* rfc2045/rfc2045.c (content_location): Plug a leak.
* unicode/unicode.c (unicode_xconvert): Plug a memory leak.
So only time will tell if the newest release with it's memory leak fix will
improve the situation at all for my server. So far it looks promising, but I still
have my fingers crossed.
I'll also note here that the memory leak refered to in the above ChangeLog
snippet has been fixed since the sqwebmail 3.5.2 production release. I just
haven't upgraded until now (3.5.3).
--
Jesse Guardiani, Systems Administrator
WingNET Internet Services,
P.O. Box 2605 // Cleveland, TN 37320-2605
423-559-LINK (v) 423-559-5145 (f)
http://www.wingnet.net
