𝕏

Invalid task

Message boards : Number crunching : Invalid task
Message board moderation

To post messages, you must log in.

AuthorMessage
DaveW

Send message
Joined: 7 Jul 22
Posts: 7
Credit: 300,007
RAC: 0
Message 1677 - Posted: 8 Jul 2022, 7:33:13 UTC

I did look to see if there was somewhere else I could post this, but a new thread seemed most appropriate.

I have had a validation error very early on ' any clues as to why:

https://denis.usj.es/denisathome/result.php?resultid=522077
ID: 1677 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
[VENETO] boboviz

Send message
Joined: 9 Apr 15
Posts: 172
Credit: 1,552,856
RAC: 0
Message 1683 - Posted: 9 Jul 2022, 19:57:05 UTC - in response to Message 1677.  

Same here, some invalid:
592424
523926
523886
Etc...
ID: 1683 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Jesús Carro
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 18 Mar 15
Posts: 284
Credit: 2,748,608
RAC: 0
Message 1688 - Posted: 11 Jul 2022, 16:37:24 UTC - in response to Message 1683.  

We detected this type of errors in the beta version and we reduced it from 5% to around 0.1%, but there still are some computers that have this problem. Most of the initial problems come from the checkpoint, but we are not sure if the small percentage that remains is still in this part of the code.

I have not checked this in the last simulations yet, but in the beta version was caused by only a few Windows computers. We will keep trying to find the cause to reduce it to 0 (or as little as possible)

Best,
Jesús.
Jesús Carro
Universidad San Jorge
@InSilicoHeart
ID: 1688 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Jesús Carro
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 18 Mar 15
Posts: 284
Credit: 2,748,608
RAC: 0
Message 1708 - Posted: 14 Jul 2022, 8:18:46 UTC

I may have found where the bug is... I'm testing it with the Beta version
Jesús Carro
Universidad San Jorge
@InSilicoHeart
ID: 1708 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
[VENETO] boboviz

Send message
Joined: 9 Apr 15
Posts: 172
Credit: 1,552,856
RAC: 0
Message 1717 - Posted: 14 Jul 2022, 21:23:50 UTC

Still some invalid wus, with 0.8 version
1073543
1073668
ID: 1717 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Jesús Carro
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 18 Mar 15
Posts: 284
Credit: 2,748,608
RAC: 0
Message 1723 - Posted: 15 Jul 2022, 6:53:58 UTC - in response to Message 1717.  

Let's check the version 0.10... I just uploaded it recently, let's see if it solves the problem
Jesús Carro
Universidad San Jorge
@InSilicoHeart
ID: 1723 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Jean-David Beyer

Send message
Joined: 6 Mar 23
Posts: 36
Credit: 2,078,354
RAC: 0
Message 2037 - Posted: 10 Mar 2023, 2:07:25 UTC

Most of my tasks complete successfully but two failed.
State: All (215) In progress (90) Validation pending (29) Validation inconclusive (0) Valid (94) Invalid (2) Error (0)
Application: All (215) Beta of DENIS-myocyte (0) Human ventricular cell models optimization (0) New human ventricular cell model (215)

Task	Work unit	Computer
7958548 3713383 	224473
7959122 3713670 	224473


Is this something I can fix, or is it at your end?
ID: 2037 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Jesús Carro
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 18 Mar 15
Posts: 284
Credit: 2,748,608
RAC: 0
Message 2038 - Posted: 10 Mar 2023, 14:40:50 UTC - in response to Message 2037.  

Hi David. It is in our side. The difference is really small, with this information I will review what is happening to try to improve. We have very few invalid simulations, to find them help us a lot.

Thank you a lot

Best,
Jesús.
Jesús Carro
Universidad San Jorge
@InSilicoHeart
ID: 2038 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Jean-David Beyer

Send message
Joined: 6 Mar 23
Posts: 36
Credit: 2,078,354
RAC: 0
Message 2040 - Posted: 11 Mar 2023, 15:28:29 UTC - in response to Message 2038.  

Hi [Jean-]David. It is in our side. The difference is really small, with this information I will review what is happening to try to improve. We have very few invalid simulations, to find them help us a lot.


Two more have failed (and 202 valid ones). So less than 1%. Do you want me to send more as they come up?

Task    Workunit Computer

8099641 3684880  224473
8099580 3733905  224473
7958548 3713383  224473
7959122 3713670  224473

ID: 2040 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Jesús Carro
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 18 Mar 15
Posts: 284
Credit: 2,748,608
RAC: 0
Message 2041 - Posted: 13 Mar 2023, 15:36:10 UTC

No, don't worry. I will check the code and later I will analyze the results of your tasks.

Thank you.
Jesús Carro
Universidad San Jorge
@InSilicoHeart
ID: 2041 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Dayle Diamond

Send message
Joined: 28 Apr 15
Posts: 18
Credit: 2,472,888
RAC: 0
Message 2065 - Posted: 13 Apr 2023, 15:38:21 UTC

I ran two Windows Updates recently, so my computer restarted twice.
I'm looking at over 50 Validation Inconclusives, with 4 tasks already marked as Invalid.

Otherwise my computer is reliable.

So something is still wrong with checkpointing. The task resumes, but the answer is wrong.
ID: 2065 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Jesús Carro
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 18 Mar 15
Posts: 284
Credit: 2,748,608
RAC: 0
Message 2068 - Posted: 14 Apr 2023, 9:25:32 UTC - in response to Message 2065.  

Hi Dayle,
We have observed an small increment in the non valid results. We are analazying what is happening. Neverless, the percentage of non valid is still very small. More than checkpointing problems, it is related usually with problems opening or saving the file (which depends on the operative system).

I will keep you updated.

Best,
Jesús.
Jesús Carro
Universidad San Jorge
@InSilicoHeart
ID: 2068 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
rsNeutrino

Send message
Joined: 12 Mar 23
Posts: 1
Credit: 4,876,889
RAC: 0
Message 2071 - Posted: 14 Apr 2023, 15:48:03 UTC - in response to Message 2068.  

Hello Jesús!
Just in case it helps with your R&D:
Like Dayle, my very reliable computer produced many invalid tasks, starting at 11th April:
https://denis.usj.es/denisathome/results.php?hostid=224703&offset=0&show_names=0&state=5&appid=

WU status as of now, filtered from this week's batch:
1050 all
16 invalid
23 inconclusive
(86 pending validation)

Also, a funky single WU with 3 invalid tasks from last week:
https://denis.usj.es/denisathome/workunit.php?wuid=4312100
ID: 2071 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
entity

Send message
Joined: 14 Apr 22
Posts: 11
Credit: 10,554,385
RAC: 0
Message 2074 - Posted: 18 Apr 2023, 16:17:14 UTC - in response to Message 2071.  

+1

I'm seeing about 1.25% invalid across my entire farm (6 systems). These are a mix of OSes (and OS levels) and processors. This is not normal for my group of machines.
ID: 2074 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
entity

Send message
Joined: 14 Apr 22
Posts: 11
Credit: 10,554,385
RAC: 0
Message 2080 - Posted: 19 Apr 2023, 18:18:17 UTC - in response to Message 2074.  

I now have 2,353 WUs pending validation and it is growing every hour. Is this normal?
ID: 2080 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Jesús Carro
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 18 Mar 15
Posts: 284
Credit: 2,748,608
RAC: 0
Message 2081 - Posted: 21 Apr 2023, 9:22:53 UTC - in response to Message 2080.  

Only in this project? It is too much.
Jesús Carro
Universidad San Jorge
@InSilicoHeart
ID: 2081 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
entity

Send message
Joined: 14 Apr 22
Posts: 11
Credit: 10,554,385
RAC: 0
Message 2082 - Posted: 21 Apr 2023, 14:19:30 UTC - in response to Message 2081.  

Most projects get to about 150 then stabilize around that point. This project reached close to 3000 pending before work ran out and it started declining. Not complaining, just curious as it seems like a lot of pending work.
ID: 2082 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote

Message boards : Number crunching : Invalid task