Ah excellent. I was a bit confused as I know the Tellapart team, which I believe you're a member of, really knows this stack really well. Hopefully you sent that back upstream :)
I still see more outages due to weird docker bugs than virtually anything else and am pretty excited to see this support land. This doesn't seem like a discussion that should really happen here though as it is the mesos direction more than Aurora's. Perhaps we should move this to the mesos users list? On Thursday, October 22, 2015, Steve Niemitz <[email protected]> wrote: > I'm in fact more familiar with them than I'd like, as I've had to modify > the docker containerizer to support both CFS and correct memory and CPU > stats. ;) > > On Thu, Oct 22, 2015 at 11:09 AM, Jeff Schroeder < > [email protected] <javascript:;> > > wrote: > > > On Thursday, October 22, 2015, Steve Niemitz > <[email protected] > > > > > wrote: > > > > > > > > > > because docker's runtime isolation (e.g., cpu, memory, etc) is not > > > > compatible with Mesos's runtime isolation > > > > > > > > > I find that kind of an odd statement, as they both just use cgroups to > > > achieve said isolation. > > > > > > I suspect you aren't familiar with the mesos isolators. Yes they both use > > control groups under the hood, but they are different. They are also > > plugins that fit in with the rest of mesos, docker does not. I'm quite > > excited about this as the docker daemon is far and above the least stable > > part of my infrastructure. > > > > These are native mesos modules which means mesos can track and manage the > > resource consumption: > > http://mesos.apache.org/documentation/latest/network-monitoring/ > > > > > > -- > > Text by Jeff, typos by iPhone > > > -- Text by Jeff, typos by iPhone
