Thanks for the answer,
I've never responded because I have no time to test recompiling _all_
libs and apache from scratch.
When I do it I'll respond.
I'm using worker because I prefer the new threaded apache model (also
I'm unsg a fastcgid and I think it's working better with the worker,
never benchmarked).
I've done a rebuild of the world 6.3 because I put the nscd backport patch.
(though I didn't moved to 7 yet, I'm waiting for 7.3 to come out).
There were only 3 files changed when I cvsup-ed 6_3 and the patch
changed some other.

You are right about the pthread - just didn't saw that it dies again.
I'll dig some more around and post if I find something.
if anyone comes up with ideas,
please respond.

cheers, valqk.

Jeremy Chadwick wrote:
> On Wed, Apr 02, 2008 at 12:25:57PM +0300, Anton - Valqk wrote:
>> ok, just made a break trough,
>> when I added cron to /etc/libmap.conf
>> (here is my libnap.conf:
>> user# cat /etc/libmap.conf
>> [httpd]
>> [cron]
>> )
>> cron stopped crashing.

> cron isn't a threaded application, so I don't see how this fixed
> anything. cron doesn't link to any threaded libraries either. Use ldd
> and see for yourself.
>> I've wrote an angry mail at sunday morning (GMT+2) about apache
>> crashing (apache-worker) after a buildworld (by ezjail-admin), and
>> added the above lines fixed my problem.

> And you didn't respond to the people who offered to help:
> You also didn't provide any details about what you upgraded from/to, or
> any other information.
> Chances are you did not rebuild all of your ports when upgrading the OS.
> The threading libraries change periodically, but do not necessarily
> change in library revision number (E.g. so.5 --> so.6).
> Also, why must you run Apache with the worker MPM? I've personally
> (meaning I'm talking about my experiences, not a general statement)
> never seen threaded Apache work -- only prefork.
> By the way, your ktrace isn't going to help much, because I don't think
> you ran ktrace with the correct arguments. I see a lot of fork()
> action, which is expected (because that's what cron does!), but no
> information about the children.
> You'd be better off using gdb on the coredump and cron, and figuring out
> where things crashed.

This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.

_______________________________________________ mailing list
To unsubscribe, send any mail to ""