| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: BOINC Agent Support
Thread: Status Inconclusive faah |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 38
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Not higher as BOINC 5.8 is quirky with work buffering and the 'slob' factor, as someone called the 1 day standard safety margin within the Work Fetch policy for the 5.8 series. No kidding. I had suspended the three faah wu's and ended up with doing only one HPF2. I resumed one faah wu and another faah wu when the HPF2 was finished. So two faah wu's were processed and one faah wu suspended. No mather what value I put in my profile Boinc refused to fetch new work. I had to explicitly resume the one faah wu left, then it finally started downloading new wu's. So now I have a bucket load on HPF2. Cheers. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
The devs are still messing around with the automatic abort mechanism. In the next client version, it should be possible for the server to abort started or unstarted results without affecting quota in any way.
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
@Didactylos: Ok the devs can provide this facility in next version. that is not the issue now..
knreed : ....Ok - it appears that batch 1593 appears to be causing each computer to return different values in the results in a very specific way (only a few values are varying). ... Variation in the results? a) Small discrepancy in the results file that does not affect the actual results and can be fixed from your side, or b) Variation in the actual outcome values which will lead to dumping those results and require a resend latter? I guess the question is: While you guys are figuring out why the variation is occuring, what do you want us to do: a) keep crunching our WU queues knowing that you will deal with the superficial variations and be able to extract what you need, or b) Would you rather that we dump the FAAH WUs now (reset project) and work on something else in the mean time (hfp2, etc...)? It is decision time gentlemen. Do you want to crunch the 159x WUs in the queues or are they a lost cause. Some one provide an answer pleeeeease. Thanks. |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Just in the Known Issues Section posted by knreed:
----------------------------------------We have talked to Scripps and the workunits for batches 1593, 1594, 1595, 1596 and 1597 contain data that is causing problems. We have only sent out batch 1593 and part of 1594. We will suspend the remaining work remaining of this set and not send it out. The next set of work that we have to send us been checked and it should be fine. We are looking at how to grant credit for all the work that has been returned. We should have a solution for that shortly. Once we have granted that credit we will be able to resume sending out work (we have to wait until credit is granted otherwise the inconclusive batches will send out additional work). [May 17, 2007 8:41:52 PM]
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
The problem is resolved: http://www.worldcommunitygrid.org/forums/wcg/viewthread?thread=13506
 The wu's which were inconclusive are validated en credit granted, cheers. I have one wu left in which I'm on error among others, the results are still pending, but I wonder if the ones on error are given credit.  |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
When credit is granted manually, it may not show up on the results status page. So don't worry too much if you can't see your credit there - I'm sure knreed awarded it correctly.
|
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Clean your farm of bad jobs in 5 minutes :O
----------------------------------------Related to aborting the bad 1593+1594 jobs, regardless of Mac or Linux clients, you need only one Windows machine with BOINCview 1.4.2 or higher to set up a multi-device filter for a particular job type, sort the work units by project, select the range(s) and hit the Abort button and Update button in 2 smooth moves to clean the queues out and tell the servers about it! Enjoy Sample screen shot with multi range selection indicated in blue and the pop up tips for the 2 buttons to operate. 
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I'd like to mention that there are still jobs sent out from the 1593-batch. Probably to validate earlier jobs who are still pending. Some jobs are pushed through with a few days notice so Boinc will do them first. I did a couple which turned out fine. One gave me a computation error. With that wu there are four more on error and two pending but I don't care that much since I know that the 1593-batch is a little problematic. I have three more 1593-wu's waiting in my queue and I will do them and see what comes out of it.
|
||
|
|
|