[Users] Automated vzctl compact triggers a kernel panic.

[Kir Kolyshkin]

On 03/23/2015 04:45 PM, Jean-Pierre Abboud wrote:
> We had an issue this weekend where one of our nodes crashed (kernel
> panic) unfortunately due to some issues connecting to DRAC console I was
> not able to capture the error.
>
> The timing of the crash coincides with a cron we've setup to run every 4
> hours to compact all containers. I know we had an older kernel (328 days
> uptime) so I'm wondering if you recommend running the compact command
> automatically and at what frequency.
>
> To add to my last email, we do have kernelcare installed so the kernel
> was patched.

Currently we are not aware of any open issues similar to yours. It 
doesn't matter
how often do you run compact, it should work without any problems. Making it
less frequent you can hide the problem but it won't be solved, so it's 
not the way
to go. Anyway, without seeing kernel oops I can't say much.

I see two options for you here:

1. Install the latest stable kernel and do a proper reboot (you can live 
migrate
containers to other servers before doing it, and migrate it back after), try
to reproduce the bug, this time catching the kernel messages, file a bug to
bugzilla.

2. Escalate it to kernelcare team -- perhaps they haven't applied or
misapplied some of our patches (this is just my speculation). As you
are paying customer with them they should take you seriously, and
I heard their attitude is very good.

Kir.