DO NOT REPLY [Bug 17717] - Zombie/Waiting states OpenBSD

[bu...@apache.org]

DO NOT REPLY TO THIS EMAIL, BUT PLEASE POST YOUR BUG 
RELATED COMMENTS THROUGH THE WEB INTERFACE AVAILABLE AT
<http://issues.apache.org/bugzilla/show_bug.cgi?id=17717>.
ANY REPLY MADE TO THIS MESSAGE WILL NOT BE COLLECTED AND 
INSERTED IN THE BUG DATABASE.

http://issues.apache.org/bugzilla/show_bug.cgi?id=17717

Zombie/Waiting states OpenBSD





------- Additional Comments From Joachim.Lutz@netlab.nec.de  2004-09-10 09:26 -------
Hello again ;-)

After some tests my student found the reason for our problem. He wants to stay 
anonymous, that's the reason I post this.

Here we go:
=======================================================================
I think the problem is IPv6 related. I didn't get zombies anymore after
deactivating listening on IPv6 sockets. 

I was able to reproduce the problem on another computer with a fresh
OpenBSD 3.5 installation. I built apache 2.0.50 from source (configure ;
make; make install) without any options. It served the default website -
no dynamic content.
The computers I tested this on were x86's. A Athlon XP and a Pentium 3.

Here are some observations I made in the logfiles (I set the LogLevel to
debug):

In the error log I have this message about once a second:
[warn] (22)Invalid argument: connect to listener
It's not there, when I start apache. It starts some time later. I could
force getting this error message by making 10-15 parallel connections.
When I do this I get:
[info] server seems busy, (you may need to increase StartServers, or
Min/MaxSpareServers), spawning 8 children, there are 0 idle, and 12
total children
and some seconds later the "connect to listener" error messages start.
Then, some hours/days later, this error messages stop again. I think it
happens when the server doesn't get any request for some hours. 
In most of the cases the zombies start coming then: When the server gets
a request it is processed correctly and logged in the access_log, but
after serving the request the child, which served the request becomes a
zombie. At some point every child has become a zombie and apache doesn't
answer to requests anymore. After a "apachectl restart" everything works
again.
Although I set the LogLevel to debug, there are no more messages in the
error_log(except at startup) than the ones mentioned above.

I hope this will help a developer to fix the problem. If you need any
additional information, feel free to contact me.
=====================================================================

I hope this will help others as well

Best Regards,
Joachim

---------------------------------------------------------------------
To unsubscribe, e-mail: bugs-unsubscribe@httpd.apache.org
For additional commands, e-mail: bugs-help@httpd.apache.org