From: Herbert Poetzl (herbert_at_13thfloor.at)
Date: Tue 14 Oct 2003 - 16:53:21 BST
On Tue, Oct 14, 2003 at 11:41:52AM -0400, Mike Savage wrote:
> Hello all,
> I hope everyone has been doing well. I have a question concerning vserver
> I have implemented vservers (2.4.20-ctx17) with multiple IP addresses
> located on separate physical networks and several of our admins have
> reported that traffic is not always moving out across the appropriate
> adapters. From various posts to the vserver mailing list, I saw that
> ctx-17a through ctx-17c fixed various networking "anomalies" :-)
> Because of driver issues, I need to stay at kernel 2.4.20, however I have
> seen that ctx-17c is only available for 2.4.21 and 2.4.22. Here is what I
> I took a virgin 2.4.21 tarball and exploded it into two separate trees. One
> tree was patched with ctx-17, while the other was patched with ctx-17c. I
> then create a patch from the diff of these two trees. Then on a 2.4.20 tree
> that had been patched with ctx-17, I applied my new ctx-17 to ctx-17c diff.
> I saw two hunks fail...one I was able to correct by hand (socket.c IIRC),
> the other I was unable to fix...it was a hunk in udp.c (IIRC) that had to do
> with ipv6, so I went on and ignored it because I do not use ipv6. My kernel
> compiled correctly, I upgraded to vserver-23 tools and upon rebooting, my
> system is up and running with 4 vservers.
> My question is this: Is what I did OK? Are the fixes that went into moving
> from ctx-17c specific to major changes to the kernel while moving from
> 2.4.20 to 2.4.21 and 2.4.22? Or are the ctx-17 to ctx-17c changes
> applicable to 2.4.20 as well?
I would suggest to use c17f, which is available for
2.4.20 too ...
get it at ...
> Thank you.