Cleaning old workunits out of the results page

Message boards : Number crunching : Cleaning old workunits out of the results page
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile Borgholio
Avatar

Send message
Joined: 2 Aug 99
Posts: 654
Credit: 18,623,738
RAC: 45
United States
Message 56554 - Posted: 22 Dec 2004, 17:42:37 UTC

I knew this question had to have been asked before, but I couldn't find it anywhere...so forgive me if I'm asking it again. Lately I've been having a little trouble configuring a pair of my slower machines, but the problems have been solved and everything is running smoothly now. However, I still have a bunch of dumped workunits listed in my results page. How long will it take for these results to be removed from the database? I'm only asking because I still see some workunits from way back in August listed on the results page, and it'd be nice for me to only see the current (or recently returned) workunits.
You will be assimilated...bunghole!

ID: 56554 · Report as offensive
Profile Benher
Volunteer developer
Volunteer tester

Send message
Joined: 25 Jul 99
Posts: 517
Credit: 465,152
RAC: 0
United States
Message 56562 - Posted: 22 Dec 2004, 18:29:29 UTC

Borg,

Results get removed by a few conditions.

1. 3 Systems have returned results and then validated.
2. More than 3 systems have returned results, but at least 3 of them are good, and results are validated.
3. After validatation the "good" result files are posted to another database.
4. If one of the servers is running the file deleter, it looks through the database for WUs that have been validated and saved. It then deletes the users uploaded result files (any other related ones).
5. After all these stages are complete, then the transitioner "might" delete the database record.

Might, because seti people have to turn on the flag that says go ahead.
A new field was added to the DB, because of earlier threads, that flags each record with the time it was last modified. This will be used to add an extra condition..."Only delete DB records if they have not been modified for ____ days", so users get a chance to look at them if they want to.


ID: 56562 · Report as offensive
Ingleside
Volunteer developer

Send message
Joined: 4 Feb 03
Posts: 1546
Credit: 15,832,022
RAC: 13
Norway
Message 56568 - Posted: 22 Dec 2004, 19:55:05 UTC

Adding a little to Benher's comments:
1 & 2: When atleast 3 "success"-results, the Validator tries to validate them. If passes validation, the wu is marked for Assimilation, if not more "results" is generated and sent out.

But, a wu have also error-limits of how many total results, or how many errors, or how many total "success" but no consensus, and if any of these limits is reached the wu is also marked for Assimilation, and any remaining results is either marked "didn't need" or "check skipped" if not already marked as "invalid".

3; The Assimilator gets passed all good & bad results, but uses a project-specific part to decide what to do with them. AFAIK in seti only the "canonical result" is being copied to the "science database". The Assimilator always sets "assimilation done" then done.

4; If a wu is marked "assimilation done", and all results accounted for, meaning either reported & tried validated or passed deadline, the File_deleter deletes all wu & result-files from upload/download-directory.

5; After File_deleter have done it's job, the db_purger (not the transitioner) can backup and remove the wu & results from the BOINC-database. For the moment it doesn't seem the db_purger have been updated to use the new entry showing record last modified, so is probably turned off till updated, or is only manually run for the moment.


Note:
The Assimilator is run exactly once per wu, either if wu validated or errored-out, but there can still be "active" results not returned and pass validation later on. This example normally always happens if target_nresults > min_quorum (like in LHC)
ID: 56568 · Report as offensive
Profile Borgholio
Avatar

Send message
Joined: 2 Aug 99
Posts: 654
Credit: 18,623,738
RAC: 45
United States
Message 56633 - Posted: 23 Dec 2004, 0:50:27 UTC - in response to Message 56568.  

Thank you both for your input. Upon closer inspection, I see that many of these old workunits only got their third result returned over the last couple weeks, so they'll probably clear out soon.
You will be assimilated...bunghole!

ID: 56633 · Report as offensive

Message boards : Number crunching : Cleaning old workunits out of the results page


 
©2024 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.