[plug] the GNOME panel that just won't die

[Craig Ringer]

James Devenish wrote:

> So, we can see that process 6442 has disappeared but 2648 has not
> acknowledged it. So, what is the state of process 2648? And is there a
> way to get `ps` to show a wider column for 'wchan', so that you can see
> it properly (that could help!)? Normally, when 2648 dies, the zombie
> process 6442 should clear up. (If it doesn't => maybe kernel bug or
> init bug?)

Problem solved - but not in the desired way. The server "went insane" at 
about 6:20, after beginning to behave more and more flakily over the 
previous couple of hours - mostly with network access. After a while it 
suddenly stopped responding over the network. 'ifdown eth2; ifup eth2' 
helped briefly, but then it stopped responding again. 'ethtool eth2' 
reported link was fine, and the interface has a statically assigned IP. 
When the machine started reporting 'no route to host' for /some/ packets 
(completely losing the rest) when pinging another machine, I decided to 
reboot it. Just as it was syncing the disks, it paniced. I would've 
liked to record the panic but I haven't yet got around to setting up the 
serial console and didn't have time to do it manually. Oh well.

Anyway, it looks like something was broken very badly, and I'd say the 
same something caused the bizarre problem with the panel that wouldn't 
die. There's probably not much point in investigating any further - and 
it's not like I can test anything now!

I'll be upgrading the kernel when I install the shiny new RAID array 
this weekend, and I'm just going to hope this is a weird kernel bug 
that's since been solved.

--
Craig Ringer