> Am 05.01.2017 um 10:04 schrieb Manfred Selz <manfred.s...@diasemi.com>: > > Hi Reuti, > > thank you for your quick reply. > Actually, when specifying the host as you suggest - via "-q *@<hostname>", > the jobs runs fine. Much obliged for this hint! > > And no, there is no reservations/backfilling/... > > I'd still like to learn why the syntax I was following does not work, but at > least I do have a good alternative now.
I also don't know, it looks like a flaw which only turns up under certain circumstances. Using "-l h=…" for serial jobs works, and also often for parallel ones. Depending on how often you need such a setup, one could also think of defining one PE per host with $pe_slots, and by selecting the PE also the machine is already chosen (by being properly attached to the host in the queue definition). -- Reuti > Regards, > Manfred > > -----Original Message----- > From: Reuti [mailto:re...@staff.uni-marburg.de] > Sent: Donnerstag, 5. Januar 2017 09:52 > To: Manfred Selz > Cc: users@gridengine.org > Subject: Re: [gridengine users] Issue with hostname specification and > parallel environment - jobs do not start > > Hi, > > Am 05.01.2017 um 07:54 schrieb Manfred Selz: > >> Hi, >> >> in my SGE 6.2u5 environment, I am seeing a strange issue when submitting >> jobs to a parallel environment while also providing a hard hostname resource >> requirement. >> This is not a standard situation, but sometimes certain benchmarks need to >> be run on one specific host only. >> >> When submitting a jobs either with a parallel environment or with a hard >> hostname resource specification, the jobs starts without delay. >> However, the combination of both sometimes keeps jobs waiting for an >> extended period of time, and I have not been able to get a clear messages >> from the "qstat -j <jobID>" report. >> >> Parallel environment settings is: >> $ qconf -sp local >> pe_name local >> slots 1000 >> user_lists NONE >> xuser_lists NONE >> start_proc_args /bin/true >> stop_proc_args /bin/true >> allocation_rule $pe_slots >> control_slaves FALSE >> job_is_first_task TRUE >> urgency_slots min >> accounting_summary TRUE >> >> The specific host being targeted has 32 slots configured for the queue being >> used, and all of them are unused at this time. >> Is anybody aware of specific issues with the combination of parallel >> environments and a hard hostname resource request? >> >> I have already tested this: >> * Removed the parallel environment request - works >> * Removed the hostname request - works >> * Removed all resource limits ("qconf -mrqs") - no change >> * Increased the "slots" limit in the PE setting - no change >> * Changed the PE allocation_rule to "round_robin" - no change > > I only saw problems when requesting a queue and a host at the same time, i.e. > "-q" & "-l h=" at the same time. The solution may work also in your case: > request the host by a queue request: > > -q "*@node123" > > >> After all, the final message in the "qstat -j <jobID>" report is always: >> cannot run in PE "local" because it only offers 0 slots > > I assume the node is free and you have no backfilling issue where slots are > reserved. > > -- Reuti > > >> >> I have seen many older reports for the "only offers 0 slots" message on >> older pages, but none specifically for the combination with a hostname spec. >> (only). >> >> Regards, >> Manfred >> >> _______________________________________________ >> users mailing list >> users@gridengine.org >> https://gridengine.org/mailman/listinfo/users > > ________________________________ > > Dialog Semiconductor GmbH > Neue Str. 95 > D-73230 Kirchheim > Managing Directors: Dr. Jalal Bagherli, Carsten Dahl > Chairman of the Supervisory Board: Rich Beyer > Commercial register: Amtsgericht Stuttgart: HRB 231181 > UST-ID-Nr. DE 811121668 > > Legal Disclaimer: This e-mail communication (and any attachment/s) is > confidential and contains proprietary information, some or all of which may > be legally privileged. It is intended solely for the use of the individual or > entity to which it is addressed. Access to this email by anyone else is > unauthorized. If you are not the intended recipient, any disclosure, copying, > distribution or any action taken or omitted to be taken in reliance on it, is > prohibited and may be unlawful. > > Please consider the environment before printing this e-mail _______________________________________________ users mailing list users@gridengine.org https://gridengine.org/mailman/listinfo/users
