Re: Completely rude alternative to HierarchicalDRF?

2016-01-25 Thread Guangya Liu
Does setting a large weight for this framework help? You can configure different frameworks use different roles. On Tue, Jan 26, 2016 at 7:18 AM, Charles Allen < charles.al...@metamarkets.com> wrote: > Is there an allocator out there that will prioritize frameworks to favor > the oldest first? >

Re: Completely rude alternative to HierarchicalDRF?

2016-01-25 Thread David Greenberg
You should check out Cook from Two Sigma. It lets your fairly share you're cluster between several spark instances, scaling them up and down based on tenure and demand. Check it out at github.com/twosigma/cook On Mon, Jan 25, 2016 at 6:18 PM Charles Allen wrote: > Is there an allocator out there

Completely rude alternative to HierarchicalDRF?

2016-01-25 Thread Charles Allen
Is there an allocator out there that will prioritize frameworks to favor the oldest first? I'm using spark in coarse mode, and I actually want the earliest registered course mode drivers to get as many resources as they want.

cgroups isolation fails with shipped example code.

2016-01-25 Thread Shiyao Ma
Hi, I built mesos from the latest master branch. Everything is fine, however, if the cgroups isolation flag is added, the the python example will fail. You can notice that the framework will never be shutdown. Detailed log is here: https://bpaste.net/show/e236858fde30 To reproduce the problem,

Re: Tasks failing when restarting slave on Mesos 0.23.1

2016-01-25 Thread Matthias Bach
Hi, Thanks for your feedback. I checked the kernel logs and they don't show anything. The Thermos logs I had already checked. They don't show any activity, as if the request never reached Thermos. Kind Regards, Matthias Am 20.01.2016 um 03:14 schrieb Benjamin Mahler: > From the slave (now known