Message boards : Theory Application : Sherpa - longest runtime with Success - native
Message board moderation

To post messages, you must log in.

AuthorMessage
maeax

Send message
Joined: 2 May 07
Posts: 914
Credit: 33,692,829
RAC: 8,158
Message 40805 - Posted: 6 Dec 2019, 6:19:25 UTC
Last modified: 6 Dec 2019, 6:19:45 UTC

[boinc pp jets 13000 180,-,3560 - sherpa 2.2.5 default 19000 190]
Laufzeit 1 Tage 10 Stunden 19 min. 59 sek. - CPU Zeit 1 Tage 10 Stunden 19 min. 22 sek.
ID: 40805 · Report as offensive     Reply Quote
Aurum
Avatar

Send message
Joined: 12 Jun 18
Posts: 80
Credit: 35,643,381
RAC: 186
Message 40811 - Posted: 6 Dec 2019, 10:48:36 UTC

Ten days, you must be a very patient person :-)
ID: 40811 · Report as offensive     Reply Quote
Luigi R.
Avatar

Send message
Joined: 7 Feb 14
Posts: 99
Credit: 5,027,000
RAC: 0
Message 40812 - Posted: 6 Dec 2019, 11:25:16 UTC

It would be nice to see runRivet.log.
ID: 40812 · Report as offensive     Reply Quote
maeax

Send message
Joined: 2 May 07
Posts: 914
Credit: 33,692,829
RAC: 8,158
Message 40814 - Posted: 6 Dec 2019, 13:02:55 UTC

[boinc pp jets 7000 800 - sherpa 2.2.2 default 34000 190]
Laufzeit 2 Tage 13 Stunden 49 min. 57 sek. - CPU Zeit 2 Tage 5 Stunden 53 min. 23 sek.
ID: 40814 · Report as offensive     Reply Quote
Henry Nebrensky

Send message
Joined: 13 Jul 05
Posts: 116
Credit: 12,845,276
RAC: 20,693
Message 40837 - Posted: 7 Dec 2019, 13:43:41 UTC - in response to Message 40814.  
Last modified: 7 Dec 2019, 14:21:45 UTC

Task 253573409: [boinc pp jets 7000 150,-,2160 - sherpa 2.2.0 default 6000 186]
Run time 4 days 3 hours 41 min 11 sec
CPU time 4 days 3 hours 10 min 24 sec

Though my record in recent times is instead a Sixtrack task (but on a slower machine):
Name workspace1_HEL_Qp_2_MO_0_1t_3s_1e7turns__5__s__62.31_60.32__1_1.5__7__80_1_sixvf_boinc21890_1
Run time 7 days 13 hours 0 min 26 sec
CPU time 7 days 12 hours 51 min 12 sec

My point is that some of us are happy to run long-lived tasks but it would help to be able to separate them out for operational reasons; hence my suggestion in another thread that the Theory-Native subproject be set aside for Sherpa 2.x tasks which are the ones that commonly over-run and need manual inspection of the log files.
ID: 40837 · Report as offensive     Reply Quote
maeax

Send message
Joined: 2 May 07
Posts: 914
Credit: 33,692,829
RAC: 8,158
Message 40839 - Posted: 7 Dec 2019, 15:38:09 UTC - in response to Message 40837.  

My point is that some of us are happy to run long-lived tasks but it would help to be able to separate them out for operational reasons; hence my suggestion in another thread that the Theory-Native subproject be set aside for Sherpa 2.x tasks which are the ones that commonly over-run and need manual inspection of the log files.

Yes Henry,
the idea is to find those successful longrunner with Sherpa 2.2.x to help us from Cern-IT or the Sherpa-Team (P.Skands).
ID: 40839 · Report as offensive     Reply Quote
maeax

Send message
Joined: 2 May 07
Posts: 914
Credit: 33,692,829
RAC: 8,158
Message 40865 - Posted: 9 Dec 2019, 9:55:08 UTC
Last modified: 9 Dec 2019, 9:56:02 UTC

[boinc ppbar ue 1800 10 - sherpa 2.2.5 default 81000 192]
Laufzeit 20 Stunden 59 min. 2 sek. CPU Zeit 20 Stunden 51 min. 9 sek.
ID: 40865 · Report as offensive     Reply Quote
Luigi R.
Avatar

Send message
Joined: 7 Feb 14
Posts: 99
Credit: 5,027,000
RAC: 0
Message 40874 - Posted: 9 Dec 2019, 15:52:00 UTC - in response to Message 40839.  

Yes Henry,
the idea is to find those successful longrunner with Sherpa 2.2.x to help us from Cern-IT or the Sherpa-Team (P.Skands).

Just get your hostid list from active hosts page on mcplots-dev.cern.ch.
Then scrape html pages from lhcathome.cern.ch looking for long tasks.
If you find someone, you check if it is sherpa on stderr output.
ID: 40874 · Report as offensive     Reply Quote
Henry Nebrensky

Send message
Joined: 13 Jul 05
Posts: 116
Credit: 12,845,276
RAC: 20,693
Message 40875 - Posted: 9 Dec 2019, 16:30:39 UTC - in response to Message 40874.  
Last modified: 9 Dec 2019, 16:31:18 UTC

Just get your hostid list from active hosts page on mcplots-dev.cern.ch. Then scrape html pages from lhcathome.cern.ch looking for long tasks.
I'm not clear how that would work - how do you distinguish a long-running task from one that's still sitting on the host unstarted?
If you find someone, you check if it is sherpa on stderr output.
I still have to search for the host and slot to find the logs. It would be much easier to only run the Sherpas on the machine where I can tolerate run times of a week or more; then it's easy to see if any Sherpa is running and for how long with top and only if it's more than say 10 hours in do I need to start poking in the logs trying to decide if it will finish or want to keep running until the heat death of the universe.
ID: 40875 · Report as offensive     Reply Quote
Luigi R.
Avatar

Send message
Joined: 7 Feb 14
Posts: 99
Credit: 5,027,000
RAC: 0
Message 40878 - Posted: 9 Dec 2019, 17:12:23 UTC - in response to Message 40875.  

I'm not clear how that would work - how do you distinguish a long-running task from one that's still sitting on the host unstarted?
We're looking for successful tasks, so I would use this page and parameters.
https://lhcathome.cern.ch/lhcathome/results.php?hostid=HOSTID&offset=0&show_names=0&state=4&appid=13

Then you select tasks by elapsed or run time.
If time > YOUR_TIME, you have a long task and go here
https://lhcathome.cern.ch/lhcathome/result.php?resultid=LONG_TASK_RESULTID
to read stderr output that contains "sherpa or not" information.
ID: 40878 · Report as offensive     Reply Quote
Luigi R.
Avatar

Send message
Joined: 7 Feb 14
Posts: 99
Credit: 5,027,000
RAC: 0
Message 40879 - Posted: 9 Dec 2019, 17:18:23 UTC - in response to Message 40875.  

I still have to search for the host and slot to find the logs.

Not if it'a a native theory task.
You can do different operations if plan class is vbox or native.
You would scan all the hosts, not only yours. ;)
ID: 40879 · Report as offensive     Reply Quote
Luigi R.
Avatar

Send message
Joined: 7 Feb 14
Posts: 99
Credit: 5,027,000
RAC: 0
Message 40905 - Posted: 11 Dec 2019, 19:50:14 UTC - in response to Message 40879.  

Successfully crunched one of those longer tasks, I guess.

boinc pp jets 7000 150,-,2160 - sherpa 2.2.5 default 2000 192

Run time: 1d9h52m35s
resultid=254968079
ID: 40905 · Report as offensive     Reply Quote
maeax

Send message
Joined: 2 May 07
Posts: 914
Credit: 33,692,829
RAC: 8,158
Message 41132 - Posted: 1 Jan 2020, 10:51:51 UTC
Last modified: 1 Jan 2020, 10:54:12 UTC

[boinc pp ttbar 7000 - - sherpa 2.1.1 default 3000 198]
Laufzeit 5 Tage 16 Stunden 47 min. 50 sek. - CPU Zeit 4 Tage 22 Stunden 16 min. 10 sek.
ID: 41132 · Report as offensive     Reply Quote
Luigi R.
Avatar

Send message
Joined: 7 Feb 14
Posts: 99
Credit: 5,027,000
RAC: 0
Message 41136 - Posted: 1 Jan 2020, 13:21:28 UTC - in response to Message 41132.  

https://lhcathome.cern.ch/lhcathome/result.php?resultid=255648161
pp jets 7000 150,-,2160 - sherpa 2.2.4 default 3000 196
runtime: 10d 13h 58m 30s
cputime: 7d 23h 19m 56s
ID: 41136 · Report as offensive     Reply Quote

Message boards : Theory Application : Sherpa - longest runtime with Success - native


©2020 CERN