Hi,
On Sun, May 10, 2009 at 02:49:47AM +0000, Andy Smith wrote:
Hi,
On Sat, May 09, 2009 at 02:19:34PM +0000, Andy Smith wrote:
This same problem happened before on this host
about 4 months ago,
but I don't believe it is a hardware issue. There is a rolling
upgrade of hypervisor and kernel being planned at the moment and I
hope that will get around this, but if not then I will replace the
hardware.
Actually thinking about it some more, I have a suspicion that this
is related to radvd and IPv6. radvd is the Router Advertisement
Daemon. It's what allows your VPSes to automatically configure
their IPv6 connectivity, and the implementation on Linux does not
seem very stable.
This same problem re-occurred today and as a result I had to hard
reboot obstler again at approximately 15:35Z today.
I didn't actually disable radvd because a number of customers
complained that they didn't know how to statically configure IPv6.
I now believe it may actually be this bug though:
http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h…
Debian Etch is indeed using 2.6.18 kernel, so I am not sure if this
is fixed in it, and Xen does shut down network devices without
removing the addresses first. I can work around this, either by
making sure the addresses are removed or else not having them in the
first place.
Also now the new servers are in I will be doing rolling upgrades of
hosts to Debian lenny amd64.
Cheers,
Andy
--
http://bitfolk.com/ -- No-nonsense VPS hosting