valterc, thank you for the work.
It seems to me that v1.10 tasks will not validate against v1.11 tasks.
Do you have any completed tasks from this batch in the database that have validated with one of each app version ?
Reply
Quote
Well a small batch is good news. It shows that someone is working the applications and that is positive after 5 months without a small batch.
____________
In October 1969 I took an oath to support and defend the Constitution of the United States against all enemies, foreign and domestic;
There was no expiration date.
Reply
Quote
Any chance for someone from the project to give an update about potential new batches?
I joined the project a couple of months ago in hopes of grabbing some of these rare tasks and now I'm getting hopeful :D
Reply
Quote
65% of tasks in yesterday's batch failed for me with
____________
I crunch for Ukraine
Reply
Quote
65% of tasks in yesterday's batch failed for me with
<core_client_version>8.2.5</core_client_version> <![CDATA[ <stderr_txt> Start @ Fri Oct 17 23:20:03 2025 [E] readCheckpoint: failed to open checkpoint at 'checkpoint.txt' |225326260| Finish @ Fri Oct 17 23:41:53 2025 23:41:53 (99649): called boinc_finish(0) </stderr_txt> ]]>
It seems that when the initial tasks get assigned to different app versions (1.10 or 1.11) they don't validate against each other so a 3rd task will get sent out and it's essentially a 50/50 chance to have your original task validated or not.
Reply
Quote
65% of tasks in yesterday's batch failed for me with
<core_client_version>8.2.5</core_client_version> <![CDATA[ <stderr_txt> Start @ Fri Oct 17 23:20:03 2025 [E] readCheckpoint: failed to open checkpoint at 'checkpoint.txt' |225326260| Finish @ Fri Oct 17 23:41:53 2025 23:41:53 (99649): called boinc_finish(0) </stderr_txt> ]]>
rilian, your problem is likely related to pausing/resuming tasks, see here...
https://gene.disi.unitn.it/test/forum_thread.php?id=339
Tasks being marked as invalid due to having a mismatch of v1.10 and v1.11 do complete successfully.
Reply
Quote
something should probably be done about the 1.10 vs 1.11 invalidations, since 1.11 is only available on windows and all other platforms Linux/Mac/Arm do not have the 1.11 app version available.
if you're not going to make 1.11 available to the other platforms, then send work in such a way that the WU is sent to hosts that will run the same app.
Reply
Quote
valtercProject administrator
Project tester
Send message
Joined: 30 Oct 13
Posts: 629
Credit: 34,725,842
RAC: 717
Well, this was completely unexpected... The application has not been changed since 2017. There is formally no difference, no source code change, between the 1.10 and 1.11 version, if I remember correctly we moved from 1.10 to 1.11 just for windows because we changed the compiling environment (gcc inside cygwin)
What we now, from the beginning, is that Linux and Windows produce slightly different results, and made some changes in order to make them equal (a cut threshold in the output list)
What was changed, our side, is the input dataset. The latest one triggered that effect again (don't know exactly why, right now).
For now there is nothing I can do apart being upset because of the waste of computational power.
There is no plan to use this dataset again, in the future we'll try to do our best in order to avoid this problem.
Reply
Quote
Thanks for the feedback!
Do you have any information about work that might be available in the not too distant future? Medical projects are all either down or out of work so we are kind of hungry for new tasks. LHC is an interesting backup, but life science projects are the ones that can quickly have positive impact on humanity :)
Reply
Quote
65% of tasks in yesterday's batch failed for me with
<core_client_version>8.2.5</core_client_version> <![CDATA[ <stderr_txt> Start @ Fri Oct 17 23:20:03 2025 [E] readCheckpoint: failed to open checkpoint at 'checkpoint.txt' |225326260| Finish @ Fri Oct 17 23:41:53 2025 23:41:53 (99649): called boinc_finish(0) </stderr_txt> ]]>
rilian, your problem is likely related to pausing/resuming tasks, see here...
https://gene.disi.unitn.it/test/forum_thread.php?id=339
Tasks being marked as invalid due to having a mismatch of v1.10 and v1.11 do complete successfully.
Hi PDW, you are probably right - i was trying to balance fan noise on a 12-cpu laptop while i work and i set 20%->100%, 100%->20% CPU usage from time to time...
Will keep this in mind next time
____________
I crunch for Ukraine
Reply
Quote