| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: Website Support
Thread: Database Maintenance - No work sent or received on Dec 13 from 16:00 UTC - 20:00 UTC [COMPLETED] |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 7
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Maybe there already was somewhere a 'discuss here' issues thread raised for this, if so I apologize for not checking, but, the good thing of this one http://www.worldcommunitygrid.org/forums/wcg/viewthread_thread,35973 compared to the full feeder scheduler off or no disk space is, that the result file upload server has continued to operate, so far, meaning if the servers are opened again, the only thing the clients need doing is fetch new work, if needed, and in the same flow clear the 'ready to report'. Then of course the validators will be doing a little overtime... to hopefully catch up by midnight.
----------------------------------------(counting 10 Mississippi 11 Mississippi, at 1:15 into the maintenance Kevin wrote in to say we were about 25% complete... that'd make it 5 hours on the abacus...could be a little later, or he started the maint job half an hour after he wrote he'd started, which could then turn into the job finishing early) Anyway, if you wish to discuss, put it below here, after these messages... http://www.youtube.com/watch?v=TIcUs-6yW78 Crunch on! [Edit 1 times, last edit by Former Member at Dec 14, 2013 10:55:23 AM] |
||
|
|
knreed
Former World Community Grid Tech Joined: Nov 8, 2004 Post Count: 4504 Status: Offline Project Badges: 
|
My estimates weren't quite linear
---------------------------------------- The system is available now.[Edit 1 times, last edit by knreed at Dec 13, 2013 7:18:59 PM] |
||
|
|
knreed
Former World Community Grid Tech Joined: Nov 8, 2004 Post Count: 4504 Status: Offline Project Badges:
|
The database likes be cleaned up from time to time:
----------------------------------------num_results yrs_runtime_returned minutes_since_returned These are the per minute results being reported back. Very few devices are getting high load messages during the recovery which is due to the database performing much better. For those who are more technically minded we ran the 'Optimize table' command against the 'Result' and 'Workunit' tables. These are two of the larger tables in the database and are subject to significant inserts and deletes. [Edit 2 times, last edit by knreed at Dec 13, 2013 9:10:04 PM] |
||
|
|
Eric_Kaiser
Veteran Cruncher Germany (Hessen) Joined: May 7, 2013 Post Count: 1047 Status: Offline Project Badges: 
|
Good work. Thanks.
---------------------------------------- |
||
|
|
cjslman
Master Cruncher Mexico Joined: Nov 23, 2004 Post Count: 2082 Status: Offline Project Badges: 
|
Great Job... it's good to see the results so soon.
----------------------------------------CJSL Crunching for a better world... |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Most impressively, measuring by stats, the afternoon numbers with the 3.5 hour outage matched the morning numbers... 266 v 265 [yet (moi) had a boat load of wingman coming in the early hours of Saturday morning. One cause could be client bounces [the deferral counter after multiple in vain retries during the outage] and the BETA 7.28 pushed ahead by testers. Still got 2 running, making progress, into their 27th hour and at about 90%. Those are holding up 2 other half way run jobs [both have a pre-assigned wingman]. They'll get their turn today :D
Kevin, when will there be a controlled time back-off sent so that clients are on first connect during an out told to stay away to at least after a certain hour [together with the maint message "maintenance underway, your client will reconnect when expected to complete in nnn minutes"]? E.g. up that 182 seconds to n hours, and decrement that value as the optimization progresses. Maybe not worth the effort for a 4 hour programmed out [think you wrote some years ago about the exponential effect the longer it takes], but if it is of interest. Crunch On. |
||
|
|
knreed
Former World Community Grid Tech Joined: Nov 8, 2004 Post Count: 4504 Status: Offline Project Badges:
|
It would be nice to add, but haven't had the opportunity to prioritize the task. At the moment the list is getting longer so it will be sometime before this can be implemented.
|
||
|
|
|