Too late to validate?

Message boards : Number crunching : Too late to validate?

To post messages, you must log in.

Bill Michael

Send message
Joined: 22 Sep 15
Posts: 8
Credit: 16,883,075
RAC: 1,612
Message 5037 - Posted: 30 May 2017, 15:32:20 UTC

Getting a lot more "failed to validate" errors (several hosts, good example 209208), but the oddest ones are "too late to validate" on AV tasks that were returned only a couple of hours after being sent out. Example: task 259182639
ID: 5037 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote

Send message
Joined: 14 Jul 09
Posts: 52
Credit: 572,365,154
RAC: 990,093
Message 5040 - Posted: 30 May 2017, 19:28:17 UTC
Last modified: 30 May 2017, 19:31:55 UTC

I found one "Validate error" result from my Enigma AV tasks. Returned 29 May 2017, 13:00:58 UTC.

Mine and yours (I didn't find task you mentioned, but I looked other failed tasks from mentioned host) weren't returned too late as there wasn't other computers processing same workunits simultanously.

Following is just shaky guess: I would guess some kind of error at server side from aftermath of server software upgrade. Only TJM can know.
ID: 5040 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile TJM
Project administrator
Project developer
Project scientist

Send message
Joined: 25 Aug 07
Posts: 843
Credit: 219,156,502
RAC: 331,309
Message 5110 - Posted: 1 Jun 2017, 11:49:40 UTC - in response to Message 5040.  

It was a bug in my code, the newer server versions use slightly changed file names for result files. The daemon that processes the results was guessing the file name by adding "_0" to the result name. This method does not work anymore and result files from tasks generated after the update could not be opened. These tasks were flagged with "internal server error" because situation where validator can open the file and the next stage can't should not happen. Eventually some of these tasks were reissued but there weren't too many of them as the server halts itself in case of many repeated failures. From client point of view this was just a minor bug as it changed workunit/result state after it was validated.

The "too late to validate" is also a bug, it was used as a workaround for old BOINC server code bug when the transitioner did not add second result to workunits with quorum=1 when the first one failed with validate state invalid.
M4 Project homepage
M4 Project wiki
ID: 5110 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote

Message boards : Number crunching : Too late to validate?

Copyright © 2019 TJM