[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [HTCondor-users] CLAIM_WORKLIFE - a couple of questions
- Date: Tue, 5 Nov 2019 19:54:32 +0000
- From: John M Knoeller <johnkn@xxxxxxxxxxx>
- Subject: Re: [HTCondor-users] CLAIM_WORKLIFE - a couple of questions
You can do SLOT_TYPE_1_CLAIM_WORKLIFE, and it can be an expression that refers to other attributes of the slot.
As for the second, there isn't anything I can think of that is per-slot or per-slot-type. RecentJobStarts is for the whole machine
and that's the closest thing I can think of.
From: HTCondor-users <htcondor-users-bounces@xxxxxxxxxxx> On Behalf Of Beyer, Christoph
Sent: Tuesday, November 5, 2019 4:45 AM
To: htcondor-users <htcondor-users@xxxxxxxxxxx>
Subject: [HTCondor-users] CLAIM_WORKLIFE - a couple of questions
I do have a coupls - OK 2 questions about the CLAIM_WORKLIFE:
- Is it possible to define the CLAIM_WORKLIFe per slot as in: 'SLOT_TYPE_1_CLAIM_WORKLIFE = 60' ?
- second = more complicated ;)
We do have occasionally jobs that fail within a second or two on a specific workernode without being recognized as a failed job by htcondor. In this case the slot attracts hundreds more of these jobs of the user over the 20 minute period time that all fail.
Hence it would be useful to zero the CLAIM_WORKLIFE in case of job-failure (maybe this is the case already ?) and it would be equally useful to compute the CLAIM_WORKLIFE more individually/dynamic like '10 times the runtime of the first job- max 1200 sec' etc. I had a short look into the condor-toolbox but it seemed difficult to realize with ship's ressources ?
Building 02b, Room 009
HTCondor-users mailing list
To unsubscribe, send a message to htcondor-users-request@xxxxxxxxxxx with a
You can also unsubscribe by visiting
The archives can be found at: