Can't report completed tasks due to "Scheduler request failed: HTTP internal server error"

Questions and Answers : Unix/Linux : Can't report completed tasks due to "Scheduler request failed: HTTP internal server error"
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile Mr. Kevvy Crowdfunding Project Donor*Special Project $250 donor
Volunteer moderator
Volunteer tester
Avatar

Send message
Joined: 15 May 99
Posts: 3165
Credit: 1,112,670,869
RAC: 930,261
Canada
Message 2001406 - Posted: 6 Jul 2019, 22:31:05 UTC
Last modified: 10 Jul 2019, 11:20:21 UTC

ID: 2001406 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 12822
Credit: 208,441,665
RAC: 68,913
Australia
Message 2001418 - Posted: 6 Jul 2019, 23:28:18 UTC

No Programme permissions relating to network connections? Firewall permissions?
Grant
Darwin NT
ID: 2001418 · Report as offensive
Profile Mr. Kevvy Crowdfunding Project Donor*Special Project $250 donor
Volunteer moderator
Volunteer tester
Avatar

Send message
Joined: 15 May 99
Posts: 3165
Credit: 1,112,670,869
RAC: 930,261
Canada
Message 2001458 - Posted: 7 Jul 2019, 2:55:02 UTC - in response to Message 2001418.  

Not that I know of... firewall is off (by default in Mint as well) and also on the other computer where this BOINC does the same, but its own connects properly.
ID: 2001458 · Report as offensive
Profile Mr. Kevvy Crowdfunding Project Donor*Special Project $250 donor
Volunteer moderator
Volunteer tester
Avatar

Send message
Joined: 15 May 99
Posts: 3165
Credit: 1,112,670,869
RAC: 930,261
Canada
Message 2001914 - Posted: 10 Jul 2019, 11:16:18 UTC
Last modified: 10 Jul 2019, 11:35:16 UTC

Well I am at a loss... the very same thing has happened again. I reinstalled the AIO on that machine wiping out all traces of the old one (backed it up to a flash drive) and it worked since the time of my last update, and now the same thing has happened again and it is refusing to contact the scheduler.

I'm adding more data into the first post as I find it so that it's all one place.

Edit: I may have found the issue in client_state.xml. If one of the video cards has issues and won't initialize, repeating error blocks are accumulating in the stderr of the work units, which are then too large to be reported so the scheduler throws an error, which in any other circumstance would mean it was down or having issues. I'm going to remove those blocks and retry. I will post the details and I'm archiving client_state.xml as this may need to go to the developers...
ID: 2001914 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 13949
Credit: 199,945,238
RAC: 206,230
United Kingdom
Message 2001920 - Posted: 10 Jul 2019, 12:20:34 UTC - in response to Message 2001406.  

I see

Sat 06 Jul 2019 07:33:08 AM EDT | SETI@home | Reporting 512 completed tasks
Please (for debugging purposes) look for the file 'sched_request_setiathome.berkeley.edu.xml' in your BOINC data directory and record the file size in KB or - more likely - MB. I have a suspicion that the internal server errors are more likely when the size of this file grows when multiple task reports are combined.

Then, find and edit your 'cc_config.xml' file. Look for the line

<max_tasks_reported>0</max_tasks_reported>
and set an actual number - I usually suggest 64. Save the file, select 'Read config files' in BOINC Manager (options menu), and try a manual update again.
ID: 2001920 · Report as offensive
Profile Mr. Kevvy Crowdfunding Project Donor*Special Project $250 donor
Volunteer moderator
Volunteer tester
Avatar

Send message
Joined: 15 May 99
Posts: 3165
Credit: 1,112,670,869
RAC: 930,261
Canada
Message 2001922 - Posted: 10 Jul 2019, 12:35:09 UTC - in response to Message 2001920.  
Last modified: 10 Jul 2019, 12:55:16 UTC

Firstly, thank you for replying!

sched_request_setiathome.berkeley.edu is 6.9MB and yup, it's full of those same repeating bad init blocks.... I'll send you a .7z of it.

cc_config.xml already has

<max_tasks_reported>100</max_tasks_reported>


It was 512 earlier. However: I had not placed it at first so if it's still showing zero, this appears to be further evidence as well as the "user initiated" scheduler request on startup that BOINC is futilely trying to complete old sched. requests rather than starting a new ones with the changed parameters.

Edit: Success! I reduced <max_tasks_reported> to 10 and it went through! Seems to also be working, albeit slowly, with 64 (nice call) so I will leave it on that permanently.
ID: 2001922 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 13949
Credit: 199,945,238
RAC: 206,230
United Kingdom
Message 2001928 - Posted: 10 Jul 2019, 13:20:08 UTC - in response to Message 2001922.  

That does seem to say that 'size matters'. Whether it's because the server runs out of memory, or because it runs out of time, I'll leave it to Eric to diagnose from the server logs.

The zero setting was a quick illustrative paste from my travelling laptop - it can't be seen remotely. I think in this case, zero means unrestricted.

To save time, remember that you can make updates as often as you like - you don't need to respect the 3:03 time out. But you DO need to respect the timeout when the backlog has cleared and you're ready to request new tasks.

BTW, I find it's also quicker to clear this sort of problem if you set 'No New Tasks' while you're clearing the backlog - turn it back on only when things are under control.
ID: 2001928 · Report as offensive

Questions and Answers : Unix/Linux : Can't report completed tasks due to "Scheduler request failed: HTTP internal server error"


 
©2020 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.