Excellent.  Thanks.

-Paul Edmon-

On 3/4/2013 4:46 PM, Andy Riebs wrote:
> Correct; "service slurm restart" (or the equivalent) across all compute
> nodes will take care of this.
>
> Andy
>
> On 03/04/2013 02:58 PM, Paul Edmon wrote:
>> Excellent thanks for letting me know.  Just for my own information I
>> would need to only restart slurm right to have this change take not
>> restart the full machine?
>>
>> -Paul Edmon-
>>
>> On 3/3/2013 10:48 PM, Andy Riebs wrote:
>>> PAul,
>>>
>>> Assuming that you are using a recent (2.5.x or later) version of SLURM,
>>> you should be able to resolve this by creating/ (or editing)
>>> /etc/sysconfig/slurm with the line
>>>
>>>         ulimit -l unlimited
>>>
>>> HTH!
>>> Andy
>>>
>>> On 03/03/2013 03:50 PM, Paul Edmon wrote:
>>>> I know this has been a topic before:
>>>>
>>>> http://comments.gmane.org/gmane.comp.distributed.slurm.devel/972
>>>>
>>>> However I also know SLURM has undergone quite a few revisions since this
>>>> time.  Still seems to be an issue though as when we set ulimit -l
>>>> ulimited in users .bashrc's those don't stick.  What is the current best
>>>> practice for solving this problem?  Ultimately it would be best to make
>>>> SLURM obey what is in the .bashrc.   This is a problem as we need to
>>>> have the locked pages set to unlimited so that the IB fabric can run
>>>> properly.  Especially since we have very large memory nodes and FDR fabric.
>>>>
>>>> Is the best way to edit the slurm start up script?  If so where should
>>>> the edit be done?  Or is it best to have this set in Grub as the last
>>>> message in the above thread suggests.
>>>>
>>>> -Paul Edmon-

Reply via email to