| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: Help Conquer Cancer
Thread: Project Help Conquer Cancer Paused Nov.16, 2007 !!!!!!! |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 35
|
|
| Author |
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Please see this post in known issues:
----------------------------------------http://www.worldcommunitygrid.org/forums/wcg/viewthread?thread=17157 Unanswered: - Will WCG submit an 'Abort' instruction to BOINC clients for any buffered work? - Does the freeze stop any redistribution of backup-copies for manually cancelled work? thanks
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
knreed
Former World Community Grid Tech Joined: Nov 8, 2004 Post Count: 4504 Status: Offline Project Badges: 
|
We are looking at our options here. Right now no work is going out for Help Conquer Cancer at all. Once we decided for sure I will post an update.
|
||
|
|
E. Frijters
Senior Cruncher The Netherlands Joined: Apr 26, 2007 Post Count: 228 Status: Offline Project Badges: 
|
I downloaded lots of HCC WU's on several machines...
----------------------------------------Can I leave them there or do you expect a change in WU-structure/data (so I have to delete them)? Good luck fixing the problem... New projects always have their suprises 
Former grid.org slave
  |
||
|
|
knreed
Former World Community Grid Tech Joined: Nov 8, 2004 Post Count: 4504 Status: Offline Project Badges:
|
Ok - the in-progress work needs to be cleared out prior to when we resume sending work for this project. In order to get this done quickly, minimize lost processor time and to allow members who have returned work to us already to be able to get credit we are doing the following:
1) We are replacing the existing validator for the project with an 'auto-validator'. This will automatically validate as valid results returned. 2) We are setting the target_nresults to 1 (it is normally 2 for Help Conquer Cancer). This number is used in the following way: After a result is returned and is in error, the validator (the state engine for BOINC) counts up how many successfully returned results, in progress results and waiting to be sent results there are. If that totals less than target_nresults, then it will generate an additional copy to be sent. Since this was 2 and now it will be one. If a member aborts a result it will not cause a new result to be sent out. 3) We are setting min_quorum to 1 (it is normally 2 for Help Conquer Cancer). This means that validation will be performed on each workunit as soon as any results are returned. Since it will automatically be set to valid the workunit will be finished. All other members assigned to the workunit will start receiving 'abort if not started' messages from the server. We will be implementing this plan over the next 15-20 minutes. |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
For completeness, the message to abort is set ready for any client that contacts the servers. The servers will not convey the message until contacted. AFAIK it only will work for clients with BOINC 5.8 and up. 5.4 will have to abort manually. The latter will simply see a message in the log, if the member happens to look.
----------------------------------------kevin, please correct me if i'm wrong. cheers
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
knreed
Former World Community Grid Tech Joined: Nov 8, 2004 Post Count: 4504 Status: Offline Project Badges:
|
Correct
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
OK, I didn't wait for the knreed's post that confirms the abort signal.
I checked my queues and found only one HCC WU on a machine running BOINC 5.10.22. I aborted that WU manually and hit the Update button. I checked the message log, there was a message that the client reported 1 task, but when I returned to the WU queue the HCC WU was still sitting there with status 'Aborted by user'. Tried update Button 2 more times, but the HCC WU remains there. In the message log every update reports again 1 completed task: 16.11.2007 20:02:39|World Community Grid|Sending scheduler request: Requested by user 16.11.2007 20:02:39|World Community Grid|Reporting 1 tasks 16.11.2007 20:03:35|World Community Grid|Scheduler RPC succeeded [server version 509] 16.11.2007 20:03:35|World Community Grid|Deferring communication for 1 min 31 sec 16.11.2007 20:03:35|World Community Grid|Reason: requested by project 16.11.2007 20:04:45|World Community Grid|Sending scheduler request: Requested by user 16.11.2007 20:04:45|World Community Grid|Reporting 1 tasks 16.11.2007 20:05:40|World Community Grid|Scheduler RPC succeeded [server version 509] 16.11.2007 20:05:40|World Community Grid|Deferring communication for 1 min 31 sec 16.11.2007 20:05:40|World Community Grid|Reason: requested by project Anything wrong going on??? Best regards Thorsten |
||
|
|
knreed
Former World Community Grid Tech Joined: Nov 8, 2004 Post Count: 4504 Status: Offline Project Badges:
|
What is the name of the workunit. I need to look at the backend.
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
It was X0000046671210200502031458_ 0-- but now a couple of minutes
and another update later the WU finally disappeared from queue and is listed on results status web page as Error. From my point of view the issue is solved. Greetings Thorsten |
||
|
|
knreed
Former World Community Grid Tech Joined: Nov 8, 2004 Post Count: 4504 Status: Offline Project Badges:
|
ok - thanks for letting us know. I believe that when we were running some queries against the db, it was not allowing your update to process (due to table locks). When your client reports work to the server, it will not remove it from your task list until it gets an acknowledgment on the reply that task was successful reported. What you described is the behavior that would occur if the ack was not returned to your client.
|
||
|
|
|