Re: [HTCondor-devel] [HTCondor-users] spontaneous reboots after enabling cgroups

[Brian Bockelman <bbockelm@xxxxxxxxxxx>]

On Jul 11, 2013, at 1:04 PM, Todd Tannenbaum <tannenba@xxxxxxxxxxx> wrote:

> 
> So... seems like we should do something about the below.  The question is what.   Some options:
> 
> 1. Never mount the freezer controller.  If we did this, does it mean that processes in a job could "get away" from our control if the job forks quickly?
> 

Yup.

> 2. Add logic in the code to not use the freezer controller if want_suspend =!= FALSE or suspend =!= false
> 

I would prefer doing this until the kernel bug is resolved.

FWIW - this doesn't seem miserably hard to reproduce, although I haven't been able to reliably reproduce.

Brian