Aaron, Glad to hear from you. I’ve heard good praises of you from everyone.
Sinfo: bart@prometheus ]# sinfo PARTITION AVAIL TIMELIMIT NODES STATE NODELIST active* up infinite 1 idle* prometheus1 active* up infinite 3 idle prometheus[2-4] debug up infinite 1 idle* prometheus6 debug up infinite 1 idle prometheus5 Dig from headnode to nodes is not resolving. Interesting too, the nodes are configured for dhcp. Thanks, Yinka. From: Aaron Knister [mailto:[email protected]] Sent: Wednesday, August 29, 2012 10:50 AM To: slurm-dev Cc: Yinka Adeosun Subject: Re: [slurm-dev] RE: sview problem, was Re: Output and Error Files Sending this again but with Yinka on the CC. My original mail is nowhere to be found :( On Wed, Aug 29, 2012 at 10:26 AM, Aaron Knister <[email protected]<mailto:[email protected]>> wrote: Hi Yinka, I've tried to send a couple e-mails this morning to the list and they didn't seem to make it so I apologize for duplicate messages. I set up the cluster you're working on and have periodically jumped in to help out over the years. Usually this problem results from communication failures between the headnode and the compute nodes. Can you send along the output of sinfo? Can you also attempt to resolve each compute node's name from the head node (ie dig @localhost bluefish1). Ensure you can resolve every compute node by name, then attempt to ping them. Best, Aaron On Wed, Aug 29, 2012 at 9:59 AM, Yinka Adeosun <[email protected]<mailto:[email protected]>> wrote: A few seconds From: Andy Riebs [mailto:[email protected]<mailto:[email protected]>] Sent: Wednesday, August 29, 2012 9:26 AM To: slurm-dev Subject: [slurm-dev] RE: sview problem, was Re: Output and Error Files On the sview problem, how long does sview report an incorrect state: a few seconds, a few minutes, until slurm is restarted? On 08/28/2012 01:24 PM, Yinka Adeosun wrote: I started here less than a week but I’d answer to the best of my knowledge. 1. What version of SLURM? (If you configured and built it, what options did you use?) – 2.1.6-1 2. What OS? What hardware? – RHEL5. Dell Clusters(HPC) 3. Are user directories shared across the cluster? --- Yes 4. Do you use automount for the user directories? -- Yes 5. Can you include a copy of your slurm.conf? --- Yes, attached 6. Do you know where the user output *is* going? –Created where the job is run 7. Do the slurmctld.log and slurmd.log log files report any errors? Yes like from nodes “…unable to register: unable to contact slurm controller (connect failure) Thanks, Yinka From: Andy Riebs [mailto:[email protected]] Sent: Tuesday, August 28, 2012 10:50 AM To: slurm-dev Subject: [slurm-dev] Re: Output and Error Files Yinka, The following information would help considerably in identifying the nature of your problem: 1. What version of SLURM? (If you configured and built it, what options did you use?) 2. What OS? What hardware? 3. Are user directories shared across the cluster? 4. Do you use automount for the user directories? 5. Can you include a copy of your slurm.conf? 6. Do you know where the user output *is* going? 7. Do the slurmctld.log and slurmd.log log files report any errors? Andy On 08/28/2012 09:55 AM, Yinka Adeosun wrote: Hi, I am having issue with slurm not creating the output(xxx.out) and error (xxx.err) file In the directory it was run. Also, sview reporting that the job is running but it’s actually not running. Yinka Adeosun Unix Administrator Vistronix, Inc Contractor to US EPA Chesapeake Bay Program Office 410-295-1323<tel:410-295-1323> -- Andy Riebs Hewlett-Packard Company High Performance Computing +1-786-263-9743<tel:%2B1-786-263-9743> My opinions are not necessarily those of HP -- Andy Riebs Hewlett-Packard Company High Performance Computing +1-786-263-9743<tel:%2B1-786-263-9743> My opinions are not necessarily those of HP
<<inline: image001.jpg>>
