[slurm-users] Requested partition configuration not available now

Mahmood Naderan mahmood.nt at gmail.com
Wed May 16 09:56:58 MDT 2018


Interesting thing I found!

As I checked the log, I saw
   part_policy_valid_acct: job's account not permitted to use this
partition (RUBY allows Y8 not y8)

However, in the command I use "-A Y8" and I am sure about that. The
parts file contains
   PartitionName=RUBY AllowAccounts=Y8 Nodes=compute-0-[2-4]

So, I decided to define y8 instead of Y8. The parts file then looks
   PartitionName=RUBY AllowAccounts=y8 Nodes=compute-0-[2-4]

and when I run "-A y8", I don't get that error.
Seems to be a bug. If there is a reason for that, please let me know.
Regards,
Mahmood




On Wed, May 16, 2018 at 3:33 PM, Mahmood Naderan <mahmood.nt at gmail.com> wrote:
> Yes they are the same.
>
> [root at rocks7 ~]# cp /etc/slurm/slurm.conf rocks7
> [root at rocks7 ~]# scp compute-0-0:/etc/slurm/slurm.conf compute-0-0
> slurm.conf
>   100% 2465     3.6MB/s   00:00
> [root at rocks7 ~]# scp compute-0-1:/etc/slurm/slurm.conf compute-0-1
> slurm.conf
>   100% 2465     4.7MB/s   00:00
> [root at rocks7 ~]# md5sum rocks7 compute-0-*
> 41df7afb1ed37cc24d8151dc8d7e6c1e  rocks7
> 41df7afb1ed37cc24d8151dc8d7e6c1e  compute-0-0
> 41df7afb1ed37cc24d8151dc8d7e6c1e  compute-0-1
>
>
>
> The cpu limit on ruby partition is 20 cores. The nodes in that
> partition are Intel Xeons with the following specs
>
> [root at rocks7 ~]# rocks run host compute-0-5 "lscpu"
> Warning: untrusted X11 forwarding setup failed: xauth key data not generated
> Architecture:          x86_64
> CPU op-mode(s):        32-bit, 64-bit
> Byte Order:            Little Endian
> CPU(s):                56
> On-line CPU(s) list:   0-55
> Thread(s) per core:    2
> Core(s) per socket:    14
> Socket(s):             2
> NUMA node(s):          2
> ...
>
> There are 14 physical cores on each cpu and therefore 28 physical
> cores which means 56 threads. The requested cores are less that 28 so,
> it should be ok. I don't know why slurm said that error.
>
>
> Regards,
> Mahmood
>
>
>
>
> On Wed, May 16, 2018 at 1:48 PM, John Hearns <hearnsj at googlemail.com> wrote:
>> Mahmood,
>> you should check that the slurm.conf files are identical on the head node
>> and the compute nodes after you run the rocks sync.
>>
>>



More information about the slurm-users mailing list