From: Herbert Poetzl (herbert_at_13thfloor.at)
Date: Mon 20 Oct 2003 - 17:13:26 BST
On Mon, Oct 20, 2003 at 10:05:46AM -0400, Matthew Nuzum wrote:
> I have two physical servers, each has at least one vserver running a
> RedHat 7.3 distribution that acts as a webserver running the standard
> RedHat apache RPMs.
> About every other Sunday morning at 4:00 am apache dies with something
> like this in the error log:
> [Sun Oct 19 04:02:02 2003] [warn] child process 20017 did not exit,
> sending another SIGHUP
> [Sun Oct 19 04:02:03 2003] [notice] SIGHUP received. Attempting to
> The only thing I can think of is that the default log rotation is
> weekly. Typically, the server load is very low on at 4:00 am, so I
> don't know why apache would die at this time.
> For some time I've been running one vserver *not as a vserver*,
> meaning that instead of using a vserver kernel, I just created an
> alias on a network card and then did "chroot /vserver/testserver" and
> started apache manually. It ran fine for months that way. Now, I've
> installed a vserver kernel and run this as a true vserver. All of a
> sudden, with absolutely NO configuration changes to the vserver
> what-so-ever it starts exhibiting this problem.
> Here is the configuration for the vserver:
> # Description: Test Server
> S_FLAGS="lock nproc"
> ULIMIT="-H -u 1000"
> Any suggestions? I do have to say, I can re-start some of the
> vservers, but there are a few that cannot be restarted unless it's
> absolutely last-resort.
I don't see any reason for apache to fail, but ...
> Host OS is RedHat 8 and Kernel is:
> Linux version 2.4.20ctx-17 (root_at_plastic.solucorp.qc.ca) (version gcc
> 3.2 20020903 (Red Hat Linux 8.0 3.2-7)) #1 dim mar 30 12:32:24 EST
please try with 2.4.22 or 2.4.23-pre7 and c17f ...
and make sure to use ULIMIT="-HS -u 1000"
Vserver mailing list