Home | About | Sematext search-lucene.com search-hadoop.com
 Search Hadoop and all its subprojects:

Switch to Plain View
HDFS, mail # user - RE: Containers and CPU


+
John Lilley 2013-07-02, 19:12
Copy link to this message
-
Re: Containers and CPU
Sandy Ryza 2013-07-02, 19:25
Use of cgroups for controlling CPU is off by default, but can be turned on
as a nodemanager configuration with
yarn.nodemanager.linux-container-executor.resources-handler.class.  So it
is site-wide.  If you want tasks to purely fight it out in the OS thread
scheduler, simply don't change from the default.

Even with cgroups on, all tasks will have access to all CPU cores.  We
don't do any pinning of tasks to cores.  If a task is requested with a
single vcore and placed on an otherwise empty machine with 8 cores, it will
have access to all 8 cores.  If 3 other tasks that requested a single vcore
are later placed on the same node, and all tasks are using as much CPU as
they can get their hands on, then each of the tasks will get 2 cores of
CPU-time.
On Tue, Jul 2, 2013 at 12:12 PM, John Lilley <[EMAIL PROTECTED]>wrote:

>  Sandy,****
>
> Sorry, I don’t completely follow.  ****
>
> When you say “with cgroups on”, is that an attribute of the AM, the
> Scheduler, or the Site/RM?  In other words is it site-wide or something
> that my application can control?****
>
> With cgroups on, is there still a way to get my desired behavior?  I’d
> really like all tasks to have access to all CPU cores and simply fight it
> out in the OS thread scheduler.****
>
> Thanks,****
>
> john****
>
> ** **
>
> *From:* Sandy Ryza [mailto:[EMAIL PROTECTED]]
> *Sent:* Tuesday, July 02, 2013 11:56 AM
> *To:* [EMAIL PROTECTED]
> *Subject:* Re: Containers and CPU****
>
> ** **
>
> CPU limits are only enforced if cgroups is turned on.  With cgroups on,
> they are only limited when there is contention, in which case tasks are
> given CPU time in proportion to the number of cores requested for/allocated
> to them.  Does that make sense?****
>
> ** **
>
> -Sandy****
>
> ** **
>
> On Tue, Jul 2, 2013 at 9:50 AM, Chuan Liu <[EMAIL PROTECTED]> wrote:*
> ***
>
> I believe this is the default behavior.****
>
> By default, only memory limit on resources is enforced.****
>
> The capacity scheduler will use DefaultResourceCalculator to compute
> resource allocation for containers by default, which also does not take CPU
> into account.****
>
>  ****
>
> -Chuan****
>
>  ****
>
> *From:* John Lilley [mailto:[EMAIL PROTECTED]]
> *Sent:* Tuesday, July 02, 2013 8:57 AM
> *To:* [EMAIL PROTECTED]
> *Subject:* Containers and CPU****
>
>  ****
>
> I have YARN tasks that benefit from multicore scaling.  However, they
> don’t **always** use more than one core.  I would like to allocate
> containers based only on memory, and let each task use as many cores as
> needed, without allocating exclusive CPU “slots” in the scheduler.  For
> example, on an 8-core node with 16GB memory, I’d like to be able to run 3
> tasks each consuming 4GB memory and each using as much CPU as they like.
> Is this the default behavior if I don’t specify CPU restrictions to the
> scheduler?****
>
> Thanks****
>
> John****
>
>  ****
>
>  ****
>
> ** **
>
+
John Lilley 2013-07-02, 19:28
+
Sandy Ryza 2013-07-02, 20:04
+
John Lilley 2013-07-02, 19:18
+
John Lilley 2013-07-02, 15:56