"Resource temporarily unavailable: setuid: unable to change to uid:" - Sarge
I've send a mail to apache-httpd-users and because I didn't found a
solution - I have to ask for help you (I suppose that it is
debian-related problem, my systems are debian sarge):
I have a strange problem with apache 1.3: it cannot fork to more then
256 processes, it exits with error:
[alert] (11)Resource temporarily unavailable: setuid: unable to change
to uid: 33
uid 33 is www-data, apache user in my system.
I've written 'strange', because:
* It is related to linux kernel 2.6 (tested on 2.6.8 on IntelXeon
(32-bit) and AMD640) - downgrade to 2.4 solves the problem.
* Probably it is not related to specific apache version (tested
1.3.26-0woody6 and 1.3.33-3 debian versions, I've found on google
someone, which had the same problem on other distribution with apache
compiled from sources)
* It is not related to PAM and limits:
server:~# su - www-data -s /bin/bash 'ulimit -u'
I've confirmed the ability to launch more then 2000 processes with a
shell test (sleep 20 2000 in background).
* It is not related to available file descriptors:
server:~# su - www-data -s /bin/bash 'ulimit -n'
server:~# cat /proc/sys/fs/file-nr
1652 387 49152
* It is related to all processes owned by one user - if I stop
apache-ssl (running as www-data too) - I can have a little bit more
* I suppose it is not related to system libraries too - I have the
problem on Debian Stable (3.0) and testing versions, there is a big
difference between them.
I'm not saying, that this is problem with apache, but I'm not able to
observe it with any other application (e.g. I have near 400 postgresql
Do you have any idea how can I find this resources limit? Maybe it has
something to do with threads?
- there is someone else, who had to downgrade to kernel 2.4
"First they ignore you. Then they laugh at you. Then they
fight you. Then you win." - Mohandas Gandhi.