[slurm-dev] Re: Invalid QOS
Hi, Thank you for your answer ! I had the good parameter in my slurm.conf, but not in scontrol show config. In fact I was only doing a scontrol reconfigure, and this parameter seems to need a restart of the daemon ! This restart solved my problem ! Thanks Antoine On 27/08/2013 17:14, Pancorbo, Juan wrote: Hi, In my case the job is rejected when submitted: *sbatch: error: Batch job submission failed: Job has invalid qos* But if I try on other clusters where the AccountingStorageEnforce is not set to qos then the job run. So I think It might be a problem with AccountingStorageEnforce. Check with */scontrol show config/* the value of AccountingStorageEnforce In the configuration that rejected the job submission I had the following value for AccountingStorageEnforce: *AccountingStorageEnforce=associations,qos* Regards Juan Pancorbo Armada juan.panco...@lrz.de mailto:juan.panco...@lrz.de http//www.lrz.de Leibniz-Rechenzentrum Abteilung: Hochleistungssysteme Boltzmannstrasse 1, 85748 Garching Telefon: +49 (0) 89 35831-8735 Fax: +49 (0) 89 35831-8535 *Von:*Chauvin Antoine [mailto:antoine.chau...@synchrotron-soleil.fr] *Gesendet:* Dienstag, 27. August 2013 16:18 *An:* slurm-dev *Betreff:* [slurm-dev] Re: Invalid QOS Hi, The job is running with the wrong qos [chauvin@isei ~]$srun --qos=runnow hostname isei103.hpc I also noticed that if i put a wrong qos, the job is running too. [chauvin@isei ~]$srun --qos=fezfezfzeggheh hostname isei103.hpc It may be for the same reason... On 27/08/2013 15:15, Aaron Knister wrote: I believe so. Does the job in question ever run or does it just sit there? If so I believe that's normal behaviour. I think there's a way to change it but I'm not sure how. Sent from my iPhone On Aug 27, 2013, at 8:18 AM, Chauvin Antoine antoine.chau...@synchrotron-soleil.fr mailto:antoine.chau...@synchrotron-soleil.fr wrote: Hi, I have this in my conf: /AccountingStorageEnforce=limits,qos/ that I believe is good, isn't it ? Antoine On 27/08/2013 14:12, Aaron Knister wrote: Hi Chauvin, Have you set the *AccountingStorageEnforce *parameter in your slurm.conf and if so what is it set to? -Aaron Sent from my iPhone On Aug 27, 2013, at 3:47 AM, Chauvin Antoine antoine.chau...@synchrotron-soleil.fr mailto:antoine.chau...@synchrotron-soleil.fr wrote: Hi, My problem is still here, I can't find where is my problem, somebody can help me ? Thanks Antoine chauvin On 08/08/2013 15:57, Chauvin Antoine wrote: Hi, I have some problem with the QOS in SLURM. I created a user chauvin like this: /sacctmgr create user chauvin Account=sr Cluster=isei DefaultAccount=sr defaultqos=parallel qoslevel=normal,mononode,parallel,urgent,heavy,serial Fairshare=parent AdminLevel=none/ /And I have in my configuration /7 QOS with one named runnow that give a plus 1 priority to the job. It is to be used only by admin. ++--+ | id | name | ++--+ | 1 | normal | | 3 | mononode | | 5 | serial | | 6 | parallel | | 7 | urgent | | 8 | runnow | | 9 | heavy| ++--+ This user chauvin should not be allowed to use this runnow qos. But when I try to do it, the job is created with this QOS. I can see the status Reason Waiting InvalidQOS for this job, but also the plus 1 priority. /JOBID USER PRIORITYAGE FAIRSHARE PARTITIONQOS 4063 chauvin 1000 1 0 1000 0 4064 chauvin 1000 1 0 1000 0 4065 chauvin 11000 1 0 1000 1 -- The job with runnow priority/ Am i doing something wrong ? Or is this something abnormal
[slurm-dev] Re: Invalid QOS
Hi, My problem is still here, I can't find where is my problem, somebody can help me ? Thanks Antoine chauvin On 08/08/2013 15:57, Chauvin Antoine wrote: Hi, I have some problem with the QOS in SLURM. I created a user chauvin like this: /sacctmgr create user chauvin Account=sr Cluster=isei DefaultAccount=sr defaultqos=parallel qoslevel=normal,mononode,parallel,urgent,heavy,serial Fairshare=parent AdminLevel=none// / /And I have in my configuration /7 QOS with one named runnow that give a plus 1 priority to the job. It is to be used only by admin. ++--+ | id | name | ++--+ | 1 | normal | | 3 | mononode | | 5 | serial | | 6 | parallel | | 7 | urgent | | 8 | runnow | | 9 | heavy| ++--+ This user chauvin should not be allowed to use this runnow qos. But when I try to do it, the job is created with this QOS. I can see the status Reason Waiting InvalidQOS for this job, but also the plus 1 priority. /JOBID USER PRIORITYAGE FAIRSHARE PARTITION QOS// //4063 chauvin 1000 1 0 1000 0// //4064 chauvin 1000 1 0 1000 0// //4065 chauvin 11000 1 0 1000 1//-- The job with runnow priority / Am i doing something wrong ? Or is this something abnormal ? Thanks -- -- Antoine CHAUVIN Groupe SR Synchrotron SOLEIL 01 69 35 93 23 -- -- Antoine CHAUVIN Groupe SR Synchrotron SOLEIL 01 69 35 93 23
[slurm-dev] Re: Invalid QOS
Hi Chauvin, Have you set the AccountingStorageEnforce parameter in your slurm.conf and if so what is it set to? -Aaron Sent from my iPhone On Aug 27, 2013, at 3:47 AM, Chauvin Antoine antoine.chau...@synchrotron-soleil.fr wrote: Hi, My problem is still here, I can't find where is my problem, somebody can help me ? Thanks Antoine chauvin On 08/08/2013 15:57, Chauvin Antoine wrote: Hi, I have some problem with the QOS in SLURM. I created a user chauvin like this: sacctmgr create user chauvin Account=sr Cluster=isei DefaultAccount=sr defaultqos=parallel qoslevel=normal,mononode,parallel,urgent,heavy,serial Fairshare=parent AdminLevel=none And I have in my configuration 7 QOS with one named runnow that give a plus 1 priority to the job. It is to be used only by admin. ++--+ | id | name | ++--+ | 1 | normal | | 3 | mononode | | 5 | serial | | 6 | parallel | | 7 | urgent | | 8 | runnow | | 9 | heavy| ++--+ This user chauvin should not be allowed to use this runnow qos. But when I try to do it, the job is created with this QOS. I can see the status Reason Waiting InvalidQOS for this job, but also the plus 1 priority. JOBID USER PRIORITYAGE FAIRSHARE PARTITIONQOS 4063 chauvin 1000 1 0 1000 0 4064 chauvin 1000 1 0 1000 0 4065 chauvin 11000 1 0 1000 1-- The job with runnow priority Am i doing something wrong ? Or is this something abnormal ? Thanks -- -- Antoine CHAUVIN Groupe SR Synchrotron SOLEIL 01 69 35 93 23 -- -- Antoine CHAUVIN Groupe SR Synchrotron SOLEIL 01 69 35 93 23
[slurm-dev] Re: Invalid QOS
Hi, I have this in my conf: /AccountingStorageEnforce=limits,qos/ that I believe is good, isn't it ? Antoine On 27/08/2013 14:12, Aaron Knister wrote: Hi Chauvin, Have you set the *AccountingStorageEnforce *parameter in your slurm.conf and if so what is it set to? -Aaron Sent from my iPhone On Aug 27, 2013, at 3:47 AM, Chauvin Antoine antoine.chau...@synchrotron-soleil.fr mailto:antoine.chau...@synchrotron-soleil.fr wrote: Hi, My problem is still here, I can't find where is my problem, somebody can help me ? Thanks Antoine chauvin On 08/08/2013 15:57, Chauvin Antoine wrote: Hi, I have some problem with the QOS in SLURM. I created a user chauvin like this: /sacctmgr create user chauvin Account=sr Cluster=isei DefaultAccount=sr defaultqos=parallel qoslevel=normal,mononode,parallel,urgent,heavy,serial Fairshare=parent AdminLevel=none// / /And I have in my configuration /7 QOS with one named runnow that give a plus 1 priority to the job. It is to be used only by admin. ++--+ | id | name | ++--+ | 1 | normal | | 3 | mononode | | 5 | serial | | 6 | parallel | | 7 | urgent | | 8 | runnow | | 9 | heavy| ++--+ This user chauvin should not be allowed to use this runnow qos. But when I try to do it, the job is created with this QOS. I can see the status Reason Waiting InvalidQOS for this job, but also the plus 1 priority. /JOBID USER PRIORITYAGE FAIRSHARE PARTITIONQOS// //4063 chauvin 1000 1 0 1000 0// //4064 chauvin 1000 1 0 1000 0// //4065 chauvin 11000 1 0 1000 1//-- The job with runnow priority / Am i doing something wrong ? Or is this something abnormal ? Thanks -- -- Antoine CHAUVIN Groupe SR Synchrotron SOLEIL 01 69 35 93 23 -- -- Antoine CHAUVIN Groupe SR Synchrotron SOLEIL 01 69 35 93 23 -- -- Antoine CHAUVIN Groupe SR Synchrotron SOLEIL 01 69 35 93 23
[slurm-dev] Re: Invalid QOS
Hi, In my case the job is rejected when submitted: sbatch: error: Batch job submission failed: Job has invalid qos But if I try on other clusters where the AccountingStorageEnforce is not set to qos then the job run. So I think It might be a problem with AccountingStorageEnforce. Check with scontrol show config the value of AccountingStorageEnforce In the configuration that rejected the job submission I had the following value for AccountingStorageEnforce: AccountingStorageEnforce=associations,qos Regards Juan Pancorbo Armada juan.panco...@lrz.demailto:juan.panco...@lrz.de http//www.lrz.de Leibniz-Rechenzentrum Abteilung: Hochleistungssysteme Boltzmannstrasse 1, 85748 Garching Telefon: +49 (0) 89 35831-8735 Fax: +49 (0) 89 35831-8535 Von: Chauvin Antoine [mailto:antoine.chau...@synchrotron-soleil.fr] Gesendet: Dienstag, 27. August 2013 16:18 An: slurm-dev Betreff: [slurm-dev] Re: Invalid QOS Hi, The job is running with the wrong qos [chauvin@isei ~]$srun --qos=runnow hostname isei103.hpc I also noticed that if i put a wrong qos, the job is running too. [chauvin@isei ~]$srun --qos=fezfezfzeggheh hostname isei103.hpc It may be for the same reason... On 27/08/2013 15:15, Aaron Knister wrote: I believe so. Does the job in question ever run or does it just sit there? If so I believe that's normal behaviour. I think there's a way to change it but I'm not sure how. Sent from my iPhone On Aug 27, 2013, at 8:18 AM, Chauvin Antoine antoine.chau...@synchrotron-soleil.frmailto:antoine.chau...@synchrotron-soleil.fr wrote: Hi, I have this in my conf: AccountingStorageEnforce=limits,qos that I believe is good, isn't it ? Antoine On 27/08/2013 14:12, Aaron Knister wrote: Hi Chauvin, Have you set the AccountingStorageEnforce parameter in your slurm.conf and if so what is it set to? -Aaron Sent from my iPhone On Aug 27, 2013, at 3:47 AM, Chauvin Antoine antoine.chau...@synchrotron-soleil.frmailto:antoine.chau...@synchrotron-soleil.fr wrote: Hi, My problem is still here, I can't find where is my problem, somebody can help me ? Thanks Antoine chauvin On 08/08/2013 15:57, Chauvin Antoine wrote: Hi, I have some problem with the QOS in SLURM. I created a user chauvin like this: sacctmgr create user chauvin Account=sr Cluster=isei DefaultAccount=sr defaultqos=parallel qoslevel=normal,mononode,parallel,urgent,heavy,serial Fairshare=parent AdminLevel=none And I have in my configuration 7 QOS with one named runnow that give a plus 1 priority to the job. It is to be used only by admin. ++--+ | id | name | ++--+ | 1 | normal | | 3 | mononode | | 5 | serial | | 6 | parallel | | 7 | urgent | | 8 | runnow | | 9 | heavy| ++--+ This user chauvin should not be allowed to use this runnow qos. But when I try to do it, the job is created with this QOS. I can see the status Reason Waiting InvalidQOS for this job, but also the plus 1 priority. JOBID USER PRIORITYAGE FAIRSHARE PARTITIONQOS 4063 chauvin 1000 1 0 1000 0 4064 chauvin 1000 1 0 1000 0 4065 chauvin 11000 1 0 1000 1-- The job with runnow priority Am i doing something wrong ? Or is this something abnormal ? Thanks -- -- Antoine CHAUVIN Groupe SR Synchrotron SOLEIL 01 69 35 93 23 -- -- Antoine CHAUVIN Groupe SR Synchrotron SOLEIL 01 69 35 93 23 -- -- Antoine CHAUVIN Groupe SR Synchrotron SOLEIL 01 69 35 93 23 -- -- Antoine CHAUVIN Groupe SR Synchrotron SOLEIL 01 69 35 93 23
