I had more than 30 WUs in a row that all stopped at 95,4% / 1:26:33 runtime. I aborted all other WU. Seems there is something wrong with this 2206.19 series. Before I had no problems.

Yesterday, I had my computer restarted, but got no new WUs until this morning. The one WU https://boinc.loda-lang.org/loda/result.php?resultid=552309 I received aborted as well at exactly 1:26:33 like the others.
WIN 10 prof, Git installed. I had no problems up to the point, when the new version 2206.19 was online.

<core_client_version>7.16.20</core_client_version>
<![CDATA[
<message>
exceeded elapsed time limit 5192.18 (86400.00G/16.64G)</message>
<stderr_txt>
10:02:39 (9452): wrapper (7.7.26016): starting
10:02:39 (9452): wrapper: running ../../projects/boinc.loda-lang.org_loda/loda-220619-windows.exe (boinc -H 2)

</stderr_txt>
]]>


Seems that "exceeded elapsed time limit 5192.18" equals 1:26:33 h. Until this problem isn't solved, it's pretty useless to try another WU.

Looks pretty much the same as on one of my machines:

<core_client_version>7.14.2</core_client_version>
<![CDATA[
<message>
exceeded elapsed time limit 6268.40 (86400.00G/13.78G)</message>
<stderr_txt>
12:52:47 (30694): wrapper (7.5.26014): starting
12:52:47 (30694): wrapper: running ../../projects/boinc.loda-lang.org_loda/loda-220619-linux-x86 (boinc -H 2)

</stderr_txt>
]]>

---

- - - - - - - - - -
Greetings, Jens

I sure have seen a lot of errors the last week or two, but slowly things seems to change for the better:
Today I made 27698 credits and that's better than any day the last week! :-)

Hope the admins are on the right track and that there will be even more improvements.
(Would also be nice to hear something from them about the scientific results obtained the last months.)
I'll stay here for a week or two more - until I get my next electricity bill...

//Gunnar

I assume that the EXIT_TIME_LIMIT_EXCEEDED error showed up again because until now we have only specified the estimated FLOPs per task, but not an upper bound / limit. It appears that BOINC calculates a bound itself based on collected stats. Therefore, we have analyzed the stats and refined the WU input settings as follows:

• Task estimate: 3.2e13 FLOPs
  
• Task bound: 1e14 FLOPs
  

The new input specification can be found here: https://github.com/loda-lang/loda-boinc/blob/master/templates/loda_in

We also published the new app version 220627 with additional bug fixes and enhancements. We'll keep monitoring the results for errors.

If you are interested in details, below are the FLOPs stats for the last 1000 tasks.
The error is thrown by BOINC here:
https://github.com/BOINC/boinc/blob/73a7754e7fd1ae3b7bf337e8dd42a7a0b42cf3d2/client/app_control.cpp#L891
I'm checking if the new settings should fix it.

2206.27 works fine, no more "time exceeded" errors. Thanks, Chris.

Since version 27 I have crunched 405 tasks of which only 11 resulted in computation error. (2.7%)
That's certainly a great improvement, and I congratulate the admins on this feat! :-)
//Gunnar

This is idiotic..I've had several tasks that surpass your so called end..

I been doing this for probably longer than you have been born but have in all those years have never encountered this error..

You were an interesting project and I realize it takes time to get up and running but honestly, I'm only in for the credits so I'll set my computer(s) to no tasks and check back sometime in the future...

Christian, the runtime limit is still not good. First is was to short but now you have gone to far the other way and it is to long.

I just had a exit time expired after 9hrs40min runtime. I asume that there was a real problem with the workunit and should have been killed faster.

So perhaps a fixed runtime limit long enough for good workunits would be beter

The error rate (timeouts) for the latest app version is 0.8%. I agree that we want to bring it down to 0.0%. Until we have all needed fixes available, I have disable the "update" profile, which causes these exceptionally long runtimes. In 24h, all newly started tasks should be ok. Please report back if you still see them on Sunday.

The stats look much better now.

There still appear to be some occurrences even after we have disabled the update profile.
I created an issue here: https://github.com/loda-lang/loda-cpp/issues/161
We have set up a long-running test. Results are posted to the #test-runtime channel in Slack.

I just got a some of these on two of my hosts, 6659 and 6653:

197 (0x000000C5) EXIT_TIME_LIMIT_EXCEEDED

Examples:
https://boinc.loda-lang.org/loda/result.php?resultid=6375446
https://boinc.loda-lang.org/loda/result.php?resultid=6369522

I'm getting hundreds of these errors on new tasks. It seems like the old (incorrect) 5 minute time limit has been put back into place and needs to be corrected.

I hope this gets fixed, until then I am not going to run LODA anymore.

We are investigating the issue.

Hi All,
the issue should be fixed in the new version 2306.03. We will monitor the new workunits. Let us know if you still see this issue with the latest app version.

Cheers,
Christian

A lot of validation errors with new version:

6467347
6467365
6467433
etc

Hi boboviz,
this is a different issue. Can you please start a new thread and check for errors in your logs?
Cheers,
Christian