[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Condor-users] When do machine RANK settings apply?





From the netogiator log snippets you posted, it appears to me that your job _was_ preempting other resource claims. What is not clear is why that same job kept coming back in subsequent negotiation cycles. Do you see anything that would explain that in the job's user log or in the ShadowLog?



Right. There's a line that says it's rejecting 94.0 and then a line that says it's preempting bchan's job for 94.0 and then back again. The ShadowLog for my submitting machine has nothing in it. The last entry is dated January 4. My ScheddLog around ~13:40 has the following:

1/5 14:38:52 Sent ad to central manager for ichesal@xxxxxxxxxx
1/5 14:38:52 Sent ad to 1 collectors for ichesal@xxxxxxxxxx
1/5 14:39:58 Activity on stashed negotiator socket
1/5 14:39:58 Negotiating for owner: ichesal@xxxxxxxxxx
1/5 14:39:58 Checking consistency running and runnable jobs
1/5 14:39:58 Tables are consistent
1/5 14:39:58 Out of jobs - 2 jobs matched, 0 jobs idle, flock level = 0
1/5 14:39:58 Sent ad to central manager for ichesal@xxxxxxxxxx
1/5 14:39:58 Sent ad to 1 collectors for ichesal@xxxxxxxxxx
1/5 14:42:28 Sent ad to central manager for ichesal@xxxxxxxxxx
1/5 14:42:28 Sent ad to 1 collectors for ichesal@xxxxxxxxxx
1/5 14:44:39 Activity on stashed negotiator socket
1/5 14:44:39 Socket activated, but could not read command
1/5 14:44:39 (Negotiator probably invalidated cached socket)
1/5 14:44:58 Sent ad to central manager for ichesal@xxxxxxxxxx
1/5 14:44:58 Sent ad to 1 collectors for ichesal@xxxxxxxxxx
1/5 14:47:28 Sent ad to central manager for ichesal@xxxxxxxxxx
1/5 14:47:28 Sent ad to 1 collectors for ichesal@xxxxxxxxxx
1/5 14:49:49 DaemonCore: Command received via TCP from host
<137.57.176.9:33313>
1/5 14:49:49 DaemonCore: received command 416 (NEGOTIATE), calling
handler (negotiate)
1/5 14:49:49 Negotiating for owner: ichesal@xxxxxxxxxx
1/5 14:49:49 Checking consistency running and runnable jobs
1/5 14:49:49 Tables are consistent
1/5 14:49:49 Out of servers - 0 jobs matched, 2 jobs idle, 0 jobs
rejected



Assuming you just have the same two jobs in your queue at all times, the above log indicates that your jobs are going from "matched" to "idle" for some reason. I would recommend turning on D_FULLDEBUG in your schedd and shadow in order to see what is happening to those jobs.

--Dan