[Recap: I and others occasionally see fork fail with EAGAIN for no
apparent reason.  In particular, we are nowhere near any limits on
the number of processed, either system wide or per user.  In my case,
this was on a busy web server that started failing non-root forks at
around 170 total processes, although root could fork at least 50 more
processes with no problem]

I've got more information on this.  The server I was seeing this problem
on was rebooted, and now it is either gone, or the limit has moved way
up.  Looking around at memory usage, number of open files, and
/proc/slabinfo, I don't *see* anything different between now and when
the machine was limiting forks.  Everything that I can think to notice
is basically the same as it was a few days ago, when the problem was
still present, except the update was around 20 days then, and is now
at 4 hours.  It sure sounds like some kind of resource leak.  This is
a 2.2.4 kernel on an SMP system.

If anyone has any suggestion as to how to track this down, feel free
to send them.  The one constraint is that I only have ssh access to
the server, and I can't install and run test kernels.

--Tim Smith


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/