\"Didn\'t need\" completed work

Message boards : Number crunching : \"Didn\'t need\" completed work
Message board moderation

To post messages, you must log in.

AuthorMessage
Fischer-Kerli
Project donor

Send message
Joined: 24 Mar 05
Posts: 70
Credit: 78,553
RAC: 0
Message 5107 - Posted: 25 Mar 2007, 17:04:17 UTC
Last modified: 25 Mar 2007, 17:20:18 UTC

I\'ve got strange messages concerning this result (WU):

2007-03-25 17:25:39 [BURP] [task_debug] result state=FILES_DOWNLOADING for 342in0.zip__ses0000000342_frm0000000261_prt00000.wu_2 from CS::update_results
2007-03-25 17:25:40 [BURP] [task_debug] result state=FILES_DOWNLOADED for 342in0.zip__ses0000000342_frm0000000261_prt00000.wu_2 from CS::update_results
...
2007-03-25 17:28:00 [BURP] Starting 342in0.zip__ses0000000342_frm0000000261_prt00000.wu_2
2007-03-25 17:28:00 [BURP] Starting task 342in0.zip__ses0000000342_frm0000000261_prt00000.wu_2 using blender version 442
2007-03-25 17:29:30 [BURP] Computation for task 342in0.zip__ses0000000342_frm0000000261_prt00000.wu_2 finished
2007-03-25 17:29:34 [BURP] [file_xfer] Started upload of file 342in0.zip__ses0000000342_frm0000000261_prt00000.wu_2_0
2007-03-25 17:30:59 [BURP] [file_xfer] Finished upload of file 342in0.zip__ses0000000342_frm0000000261_prt00000.wu_2_0
2007-03-25 17:30:59 [BURP] [file_xfer] Throughput 1549 bytes/sec
...
2007-03-25 18:21:15 [BURP] Sending scheduler request: Requested by user
2007-03-25 18:21:15 [BURP] Reporting 18 tasks
2007-03-25 18:21:25 [BURP] Scheduler RPC succeeded [server version 509]
25.03.2007 18:21:25|BURP|Message from server: Completed result 342in0.zip__ses0000000342_frm0000000261_prt00000.wu_2 refused: this result wasn\'t sent (not needed)

From the \"explain\" link on the WU page: \"Didn\'t need: The result wasn\'t sent to a client because enough other results were returned for this work unit\". In this case however, the result HAS BEEN sent (and processed). It doesn\'t bother me much because only some seconds of CPU time have been lost, but I think we have a scheduler bug here with the potential of more serious problems. I don\'t know if this is BURP specific, but I\'ve only seen it on this project. (I\'ve noticed other \"Didn\'t need\" results on BURP WU pages before with a filled-in green deadline instead of the usual \"---\".)
ID: 5107 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Janus
Volunteer moderator
Project administrator
Avatar

Send message
Joined: 16 Jun 04
Posts: 4574
Credit: 2,100,463
RAC: 8
Message 5108 - Posted: 25 Mar 2007, 17:43:41 UTC
Last modified: 25 Mar 2007, 17:43:57 UTC

Hm
ID: 5108 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Fischer-Kerli
Project donor

Send message
Joined: 24 Mar 05
Posts: 70
Credit: 78,553
RAC: 0
Message 5110 - Posted: 25 Mar 2007, 17:47:41 UTC - in response to Message 5108.  

Hm


That\'s what I thought. BTW: Here\'s how a WU with unneeded results should look like.
ID: 5110 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile AndyK
Project donor
Avatar

Send message
Joined: 2 Apr 05
Posts: 137
Credit: 20,063
RAC: 0
Message 5124 - Posted: 27 Mar 2007, 10:27:38 UTC

got it, too:
	Host	Project	Date	ID	Message
	voyager	BURP	26.03.2007 21:09:26	148	Starting 340in0.zip__ses0000000340_frm0000000220_prt00035.wu_3
	voyager	BURP	26.03.2007 21:31:00	159	Computation for task 340in0.zip__ses0000000340_frm0000000220_prt00035.wu_3 finished
	voyager	BURP	26.03.2007 21:31:02	162	[file_xfer] Started upload of file 340in0.zip__ses0000000340_frm0000000220_prt00035.wu_3_0
	voyager	BURP	26.03.2007 21:31:08	163	[file_xfer] Finished upload of file 340in0.zip__ses0000000340_frm0000000220_prt00035.wu_3_0
	voyager	BURP	26.03.2007 21:42:49	178	Message from server: Completed result 340in0.zip__ses0000000340_frm0000000220_prt00035.wu_3 refused: this result wasn\'t sent (not needed)

ID: 5124 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Fischer-Kerli
Project donor

Send message
Joined: 24 Mar 05
Posts: 70
Credit: 78,553
RAC: 0
Message 5125 - Posted: 27 Mar 2007, 10:43:53 UTC

Completed result refused: this result wasn\'t sent


Franz Kafka? Is it you?
ID: 5125 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Zanthius
Project donor

Send message
Joined: 24 Mar 05
Posts: 94
Credit: 1,627,664
RAC: 0
Message 5127 - Posted: 27 Mar 2007, 18:13:47 UTC

Looked through the logs... me too

27/03/2007 2:21:36 AM|BURP|Reporting 1 tasks
27/03/2007 2:21:41 AM|BURP|Scheduler RPC succeeded [server version 509]
27/03/2007 2:21:41 AM|BURP|Message from server: Completed result 340in0.zip__ses0000000340_frm0000000211_prt00018.wu_4 refused: this result wasn\'t sent (not needed)
27/03/2007 2:22:30 AM|BURP|Computation for task 340in0.zip__ses0000000340_frm0000000211_prt00050.wu_4 finished
27/03/2007 2:22:30 AM|BURP|Starting 340in0.zip__ses0000000340_frm0000000216_prt00018.wu_2
27/03/2007 2:22:30 AM|BURP|Starting task 340in0.zip__ses0000000340_frm0000000216_prt00018.wu_2 using blender version 442
27/03/2007 2:22:32 AM|BURP|[file_xfer] Started upload of file 340in0.zip__ses0000000340_frm0000000211_prt00050.wu_4_0
27/03/2007 2:22:36 AM|BURP|[file_xfer] Finished upload of file 340in0.zip__ses0000000340_frm0000000211_prt00050.wu_4_0
27/03/2007 2:22:36 AM|BURP|[file_xfer] Throughput 5639 bytes/sec
27/03/2007 2:22:38 AM|BURP|Sending scheduler request: To report completed tasks
27/03/2007 2:22:38 AM|BURP|Reporting 1 tasks
27/03/2007 2:22:43 AM|BURP|Scheduler RPC succeeded [server version 509]
27/03/2007 2:22:43 AM|BURP|Message from server: Completed result 340in0.zip__ses0000000340_frm0000000211_prt00050.wu_4 refused: this result wasn\'t sent (not needed)

ID: 5127 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
axarydax

Send message
Joined: 8 Mar 07
Posts: 11
Credit: 1,604
RAC: 0
Message 5145 - Posted: 28 Mar 2007, 13:36:59 UTC

isn\'t it because the workunit is already completed and validated, so it does not need other results from that WU?
ID: 5145 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Fischer-Kerli
Project donor

Send message
Joined: 24 Mar 05
Posts: 70
Credit: 78,553
RAC: 0
Message 5244 - Posted: 2 Apr 2007, 1:36:11 UTC - in response to Message 5145.  

isn\'t it because the workunit is already completed and validated, so it does not need other results from that WU?


In this case, it doesn\'t NEED them (so unsent results will never be sent => \"didn\'t need\"), but it should nevertheless ACCEPT results that have already been sent. That\'s what this is all about.
ID: 5244 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Starshine

Send message
Joined: 25 Aug 05
Posts: 4
Credit: 4,485
RAC: 0
Message 5361 - Posted: 12 Apr 2007, 19:30:51 UTC

Any way I feel that credit should be granted even for those that
you get \'didn\'t need\'. Work has been done and is correct.
ID: 5361 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Janus
Volunteer moderator
Project administrator
Avatar

Send message
Joined: 16 Jun 04
Posts: 4574
Credit: 2,100,463
RAC: 8
Message 5367 - Posted: 12 Apr 2007, 20:50:23 UTC
Last modified: 12 Apr 2007, 20:51:16 UTC

Yes it\'s a bug - it isn\'t supposed to reject them at all.
I\'m still trying to figure out how the results can be sent and not sent at the same time. It sounds like a BOINC scheduler bug to me.
ID: 5367 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
PovAddict
Avatar

Send message
Joined: 25 Apr 05
Posts: 347
Credit: 4,618
RAC: 0
Message 5371 - Posted: 13 Apr 2007, 1:41:59 UTC

When did you last upgrade the BOINC server software?
ID: 5371 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Janus
Volunteer moderator
Project administrator
Avatar

Send message
Joined: 16 Jun 04
Posts: 4574
Credit: 2,100,463
RAC: 8
Message 5378 - Posted: 13 Apr 2007, 7:48:22 UTC

Alpha switchover
ID: 5378 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile AndyK
Project donor
Avatar

Send message
Joined: 2 Apr 05
Posts: 137
Credit: 20,063
RAC: 0
Message 5396 - Posted: 13 Apr 2007, 22:02:54 UTC

ID: 5396 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Janus
Volunteer moderator
Project administrator
Avatar

Send message
Joined: 16 Jun 04
Posts: 4574
Credit: 2,100,463
RAC: 8
Message 5399 - Posted: 13 Apr 2007, 22:40:49 UTC

I\'m still looking for some kind of commonallity between all these workunits - so far I\'ve got nothing.
ID: 5399 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Achim

Send message
Joined: 17 May 05
Posts: 183
Credit: 2,642,713
RAC: 0
Message 5410 - Posted: 14 Apr 2007, 6:15:18 UTC

At least they are all send after the minimum quorum of valid results was reported.
(I checked the first 4).

Maybe the validator starts validating(when the minimum is reported), maybe the validator remembers the WU state??.

Now a client requests work, and get some of the unsend.

Now the validator finishes, and sets all which where unsent when it started to didn\'t need.
ID: 5410 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
FreeLarry

Send message
Joined: 10 Oct 04
Posts: 42
Credit: 1,689,701
RAC: 0
Message 5705 - Posted: 1 May 2007, 7:11:25 UTC

Looks like i have one orphaned result from session 377. marked as not needed on wu results but claimed and initial status on result http://burp.boinc.dk/result.php?resultid=1813122

Not enough credit to be upset but a little tweaking looks to be in order to prevent happening to many times in production runs. ;)

Larry
ID: 5705 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Janus
Volunteer moderator
Project administrator
Avatar

Send message
Joined: 16 Jun 04
Posts: 4574
Credit: 2,100,463
RAC: 8
Message 5706 - Posted: 1 May 2007, 7:41:23 UTC
Last modified: 1 May 2007, 7:42:23 UTC

Has anyone seen this happen on sessions with number greater than or equal to 380?
(I\'m trying out different scheduler settings in order to nail down this elusive bug)
ID: 5706 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Thamir

Send message
Joined: 12 Jun 06
Posts: 76
Credit: 114,295
RAC: 0
Message 5707 - Posted: 1 May 2007, 8:13:54 UTC - in response to Message 5706.  

Has anyone seen this happen on sessions with number greater than or equal to 380?
(I\'m trying out different scheduler settings in order to nail down this elusive bug)


Only got one!


Result ID
click for details Computer Sent Time reported
or deadline
explain Server state
explain Outcome
explain Client state
explain CPU time (sec) claimed credit granted credit
1940731 28163 28 Apr 2007 3:26:08 UTC 28 Apr 2007 3:37:46 UTC Over Success Done 18.28 0.06 0.07
1940732 27211 28 Apr 2007 3:31:27 UTC 28 Apr 2007 3:32:37 UTC Over Success Done 19.16 0.08 0.07
1940733 15233 28 Apr 2007 3:47:47 UTC 29 Apr 2007 7:34:27 UTC Over Didn\'t need New 0.00 --- ---
1940734 18988 28 Apr 2007 3:22:56 UTC 28 Apr 2007 3:29:13 UTC Over Success Done 34.91 0.07 0.07
ID: 5707 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile AndyK
Project donor
Avatar

Send message
Joined: 2 Apr 05
Posts: 137
Credit: 20,063
RAC: 0
Message 5713 - Posted: 1 May 2007, 11:56:58 UTC - in response to Message 5706.  

Has anyone seen this happen on sessions with number greater than or equal to 380?
(I\'m trying out different scheduler settings in order to nail down this elusive bug)


No, the latest one I got was with session 372: result 1817220
ID: 5713 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote

Message boards : Number crunching : \"Didn\'t need\" completed work