From: Liam Helmer (linuxlists_at_thevenue.org)
Date: Fri 23 Jan 2004 - 00:47:45 GMT
Now everyone's going to think that I spend all my time talking to
OK, I've solved the apache problem: lack of random pool! I left an
apache process to start... it took a long time, but it did start
eventually (notice it took 13 minutes to generate the last secret!):
[Fri Jan 23 00:27:04 2004] [notice] Digest: generating secret for digest
[Fri Jan 23 00:28:10 2004] [notice] Digest: generating secret for digest
[Fri Jan 23 00:29:16 2004] [notice] Digest: generating secret for digest
[Fri Jan 23 00:30:22 2004] [notice] Digest: generating secret for digest
[Fri Jan 23 00:43:17 2004] [notice] Digest: done
OK... I just have to get it more random data -> that, I can deal with!
On Thu, 2004-01-22 at 15:28, Liam Helmer wrote:
> Hey guys,
> I've been working with linux vservers inside my distribution
> (http://desktopappliances.org). I'm currently playing with a web server
> module running in a vserver. However, I have 2 bugs that are cropping
> up, and I'm trying to identify and correct them.
> First off, the obligatory setup info:
> Distro: Gentoo/PEA
> fs: tmpfs, reiserfs and squashfs filesystems in use, reiserfs running on
> lvm volumes
> Kernel: 2.4.24, vs1.2.2, with openwall patches and a few others (my own
> Hardware: cobalt raq3
> The first bug I think I saw a reference to here before a week or two
> ago, (although it looked like it was outside the vserver context?): If I
> have more than 1 ip address selected in the vserver context (chbind --ip
> x --ip y), then I'm able to make tcp connections to remote hosts from
> within the vserver context, but not udp connections (udp connections
> give me an error "socket.c: x.x.x.x#y invalid argument"). This is solved
> by the workaround of having the first ip being used be on the interface
> where the default route exists.
> The second bug is a little more tricky. The first time that I invoke
> apache, it starts up fine, with no issues. Then, after a random number
> of restarts, I seem to have a problem with apache -> it will no longer
> properly restart -> It hangs during the startup -> It creates a
> listener, but fails to fork off additional apache processes. I'm not
> seeing any errors, however, and have yet to glean anything from an
> strace. I can't find any particular activity that triggers it... it just
> happens, at some point after the first startup. Once it's gotten into
> this state, it never starts up correctly again -> the initial listener
> on port 80 is there, but, as the apache child processes aren't spawning,
> I get no actual web activity.
> My troubleshooting, at this point, has been to recompile apache with the
> exact kernel version, etc, in order to make sure that it's all in sync.
> My next step is going to be trying this all on another box, and setting
> up some kind of continuous strace on apache.
> Has anyone experienced anything similar? Is there anything I should look
> By the way, if anyone's curious, I've been successful in getting
> XFree86/KDE/Gnome running in a vserver <g>... there's been a couple of
> minor issues, but I think I've licked most of them.
> Vserver mailing list
Vserver mailing list