All WUs canceled by project

Message boards : Number crunching : All WUs canceled by project
Message board moderation

To post messages, you must log in.

AuthorMessage
Johannes Elsner

Send message
Joined: 14 Apr 18
Posts: 5
Credit: 8,065,904
RAC: 140
Message 441 - Posted: 18 Jun 2018, 0:10:56 UTC

Hey Towarisch,

that's pretty cool: all WUs canceled by project!
(18.06.2018 01:00 MEST) "Result is no longer usable".
Is this ok or will it be a really big fat black cat this morning? (;-))))
Greetz

JoE from germany
ID: 441 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
StyM

Send message
Joined: 26 Jan 18
Posts: 2
Credit: 16,579,090
RAC: 0
Message 442 - Posted: 18 Jun 2018, 1:01:51 UTC

same here 3k+ cancelled wu's
ID: 442 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Putin Has a Small Dick

Send message
Joined: 12 Jun 18
Posts: 14
Credit: 20,011,948
RAC: 1,302
Message 443 - Posted: 18 Jun 2018, 3:01:02 UTC

Yep, same here. Shitload cancelled as well. WTF?

At least make an announcement so us users know what is going on and why.
ID: 443 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
hoarfrost
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Help desk expert

Send message
Joined: 11 Aug 17
Posts: 644
Credit: 22,389,911
RAC: 12,496
Message 444 - Posted: 18 Jun 2018, 3:10:59 UTC
Last modified: 18 Jun 2018, 7:22:12 UTC

Hello folks!

Looks strange. Tonight I deleted a bunch of old tasks and now check for errors.
All tasks on my computers - crashed also. ;D

P.S. And thank you for information about problem!
ID: 444 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
hoarfrost
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Help desk expert

Send message
Joined: 11 Aug 17
Posts: 644
Credit: 22,389,911
RAC: 12,496
Message 445 - Posted: 18 Jun 2018, 3:23:58 UTC

Generation of new tasks - restored and I continue the analysis of case.
ID: 445 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile PDW

Send message
Joined: 8 Sep 17
Posts: 34
Credit: 100,058,938
RAC: 8
Message 446 - Posted: 18 Jun 2018, 8:17:30 UTC - in response to Message 445.  

Hi,

Having checked this morning I found that most (all ?) of my machines didn't get their tasks cancelled overnight as they had backed off unable to contact the server so they continued to run and process what was in their caches.

I've got just over 2.1k of these older WUs sitting in pending validation, the task details show successful completion but clicking on work unit details it says "Unable to handle request, can't find workunit".

I've stopped work for now, I'll wait for you to report back.
ID: 446 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
hoarfrost
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Help desk expert

Send message
Joined: 11 Aug 17
Posts: 644
Credit: 22,389,911
RAC: 12,496
Message 447 - Posted: 18 Jun 2018, 9:56:08 UTC

PWD, hello!

It's very intresting also! if your hosts saved the computed results - we have a chance to return them into project! "Manual work" in database is required for it, but if it is successful - it will be a very nice!
ID: 447 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile PDW

Send message
Joined: 8 Sep 17
Posts: 34
Credit: 100,058,938
RAC: 8
Message 448 - Posted: 18 Jun 2018, 12:04:10 UTC - in response to Message 447.  

I think others may have done the same as mine did, if you can work your magic and get all the affected pending validation tasks for everybody back into the database that would be good :-)
ID: 448 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Putin Has a Small Dick

Send message
Joined: 12 Jun 18
Posts: 14
Credit: 20,011,948
RAC: 1,302
Message 449 - Posted: 18 Jun 2018, 22:22:11 UTC

Yep, this is getting old. Now more "Aborted by Project" WUs.

Let's get this stuff in order please. Thanks!
ID: 449 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
hoarfrost
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Help desk expert

Send message
Joined: 11 Aug 17
Posts: 644
Credit: 22,389,911
RAC: 12,496
Message 450 - Posted: 19 Jun 2018, 5:02:49 UTC
Last modified: 19 Jun 2018, 5:22:12 UTC

I recreated information about workunits that does not have a canonical result - with results, that was in "In progress" state. As I see - some of my results, that was in "Wait for validation state" passed a validation and get a credit. Please try to report results to project.

Information about completed workunits (with canonical result) I try to remaster today evening.

If result became "Cancelled" it lost for processing, because no any data from it - both on BOINC client and server.

Thank you for participation! :)
ID: 450 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile PDW

Send message
Joined: 8 Sep 17
Posts: 34
Credit: 100,058,938
RAC: 8
Message 451 - Posted: 19 Jun 2018, 6:58:13 UTC - in response to Message 450.  

Thanks hoarfrost, it is looking good :)

In my account on the website I can see:

1) a lot of WUs have been validated as you have matched up the already returned task results with each other (from other users).

2) that some tasks have been recreated, they appear as in progress on the website but are not on the machine (and don't get downloaded), these are like ghosts and I expect will time out eventually and be re-issued.

3) tasks that are still pending have a cancelled task associated with it but not yet had a third task sent out for processing to validate them. I expect these will eventually be issued when more tasks are generated.

I did have some running tasks on a few machines cancelled at the time so they were aborted and lost. Fortunately most of my machines did not get the work cancelled and did report the completed results.

Some of my newer in progress tasks must have been cancelled (like bluestang reported) but they were all suspended anyway so didn't affect me as such.

Thanks again for the time and effort in getting this fixed.
ID: 451 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
hoarfrost
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Help desk expert

Send message
Joined: 11 Aug 17
Posts: 644
Credit: 22,389,911
RAC: 12,496
Message 452 - Posted: 19 Jun 2018, 7:25:35 UTC

PDW, thank you for report about tasks state! This is very useful information due to significantly bigger space of tasks on your machines!

Depending of combination of BOINC server and client versions, ghosts task may be resend on computer or marked as cancelled on server or simply stay "In progress" state util deadline. That is absolutely normal situation.

Due to marking a big set of results as "Cancelled" for many workunits needs an additional task, that was be created and pushed into queue. Now queue size is 90908 tasks, consequently, after 2 or 3 days (depending of processing power of participants computers) newly created tasks will be send to computers of participants and become "In progress" instead of "Unsent".
ID: 452 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile PDW

Send message
Joined: 8 Sep 17
Posts: 34
Credit: 100,058,938
RAC: 8
Message 453 - Posted: 19 Jun 2018, 7:51:06 UTC - in response to Message 452.  

Thanks for the confirmation about what will happen with the tasks.

I will start moving back today to work through the task queue and get it down, there are going to be three weeks of Formula Boinc sprints starting this week so you might get picked for one of those, that will get the queue down !
ID: 453 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
hoarfrost
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Help desk expert

Send message
Joined: 11 Aug 17
Posts: 644
Credit: 22,389,911
RAC: 12,496
Message 454 - Posted: 21 Jun 2018, 5:20:12 UTC
Last modified: 21 Jun 2018, 12:51:37 UTC

PDW, very nice news! Restoring data about completed workunits required more work but now about 1/3 information is restored. If any slowdown of server will be noticed, please report about it here. This restoring, I think, will be complete in 1-2 days.

And thank to all participants! Happy crunching!

P.S. Information about workunits - restored.
ID: 454 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote

Message boards : Number crunching : All WUs canceled by project

©2024 The searchers team, Karelian Research Center of the Russian Academy of Sciences