climateprediction.net home page
Lost trickles?

Lost trickles?

Questions and Answers : Macintosh : Lost trickles?
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile old_user216062

Send message
Joined: 29 Dec 06
Posts: 17
Credit: 379,624
RAC: 0
Message 32791 - Posted: 29 Feb 2008, 11:59:05 UTC
Last modified: 29 Feb 2008, 12:02:08 UTC

I seem to have a gap in trickles between 25 Feb. and 29 Feb. for workunit 7259211.

My messages tab shows 2 trickles on 26 Feb., a trickle and file upload on 27 Feb., and another trickle on 28 Feb. In each case, I got the \"Scheduler request succeeded: got 0 new tasks\" message.\"

[edit] The trickles are missing from the results page and the credits have not changed since 25 Feb. Any ideas what is wrong?

Thanks.
ID: 32791 · Report as offensive     Reply Quote
Profile Iain Inglis

Send message
Joined: 9 Jan 07
Posts: 467
Credit: 14,549,176
RAC: 317
Message 32792 - Posted: 29 Feb 2008, 13:28:34 UTC

Some of the Web site servers were taken down for maintenance earlier this week and the trickle display part of the system has not yet restarted. The trickles are being accepted: it\'s just a display and credit issue - they\'ll turn up in the end.

A message will be put in the News thread when there\'s a change. It\'s a good idea to subscribe to that thread.
ID: 32792 · Report as offensive     Reply Quote
Profile old_user216062

Send message
Joined: 29 Dec 06
Posts: 17
Credit: 379,624
RAC: 0
Message 34123 - Posted: 22 Jun 2008, 12:17:52 UTC

It\'s happened again. I\'m missing trickles between 17 June and 21 June for task 7491271. While my messages tab shows 12 trickles between 17 June and 21 June, none show up in my results page. There is also a jump in the ave sec/TS from 1.5881 for the last trickle on 17 June to 3.1740 for the first trickle on 21 June. However, my messages tab show a consistent 3 hours of wall clock time between trickles.

Any ideas?

Thanks.
ID: 34123 · Report as offensive     Reply Quote
Profile Iain Inglis

Send message
Joined: 9 Jan 07
Posts: 467
Credit: 14,549,176
RAC: 317
Message 34124 - Posted: 22 Jun 2008, 13:09:51 UTC

This has a different cause. Slab models - i.e. HADSM3 - will sometimes respond to problems by reverting to the start of a phase. This usually happens at the end of a phase if the end-of-phase post-processing is interrupted. However, your model seems to have had a problem between timestep 172,832 and 183,634 - at which point it decided to re-do the whole phase, so the CPU time doubled by the time it reached 183,634.

It should now carry on as normal.
ID: 34124 · Report as offensive     Reply Quote

Questions and Answers : Macintosh : Lost trickles?

©2024 cpdn.org