| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: BOINC Agent Support
Thread: BOINC: v7 Minimum work buffer / Max additional work buffer & v6 Connect about every xx days / Additional work buffer operation |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 40
|
|
| Author |
|
|
Ingleside
Veteran Cruncher Norway Joined: Nov 19, 2005 Post Count: 974 Status: Offline Project Badges: 
|
Away from the aside, the main being that once used to thinking minimum / maximum [cache], it's almost like "why did the developers not do this long ago ;?". work_buffer_max_days was removed with the release of v4.  "I make so many mistakes. But then just think of all the mistakes I don't make, although I might." |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Different page of the manual I guess. Minimum in the new order + the value of the Maximum additional cache gives Maximum overall cache.
 --//-- |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
OK, with some 2 weeks more of testing and observing, the draft FAQ still reads as things operate. Have tweaked the text a bit for clarity and decided on giving the Minimum work buffer and Max additional work buffer the short monikers of MinB and MaxAB. Clearly MinB with a MaxAB of zero does the old backfill work requesting trickles [when connected to the internet]. MaxAB is an extra but wont be acted on at all until the cache hits the MinB value. This also makes things true to actual operation, and we are after all most of the time connected permanently to the internet, even though the client network schedule may be set to limited hours.
------------------------------------------//-- edit: And we have a very long topic title ;P [Edit 1 times, last edit by Former Member at Feb 21, 2012 11:00:23 AM] |
||
|
|
nanoprobe
Master Cruncher Classified Joined: Aug 29, 2008 Post Count: 2998 Status: Offline Project Badges: 
|
In case it escaped, I edited into an earlier post that 7.0.12-7.0.14 alpha are thoroughly unfit for running at WCG. They mangle .gzb files [seeing odd posts on our forums suggesting more are walking into this wall]. Must be specific to certain setups. I'm running 7.0.14 on 3 machines running WCG and GPU projects together and WCG runs fine. These are all 64 bit machines so that could be a possible explanation.
In 1969 I took an oath to defend and protect the U S Constitution against all enemies, both foreign and Domestic. There was no expiration date.
----------------------------------------  [Edit 2 times, last edit by nanoprobe at Feb 21, 2012 2:19:17 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
"Must be..." No, it's just that you have not come across the sciences that have these .gzb/.gz files (latter of HPF2 failing in 7.0.15 and said to be fixed in alpha 7.0.17). Confirmed, reproducible on random test machines and fixed at developers after WCG tech involvement ;>)
--//-- |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Have tweaked the text a bit for clarity That's good, it's very clear.Minor: will resume it's old method should be "its". Minor: A sample how a cache builds ... [Now called Maximum "additional" work buffer", running only GFAM should be buffer], And we have a very long topic title ;P Purely for symmetry between v7 and v6, days and Additional should be days / Additional. There, it's now shorter  |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Let me call teacher, to leave me kids alone ;>)
(thanks for proof reading) --//-- |
||
|
|
nanoprobe
Master Cruncher Classified Joined: Aug 29, 2008 Post Count: 2998 Status: Offline Project Badges:
|
"Must be..." No, it's just that you have not come across the sciences that have these .gzb/.gz files involvement ;>) That could be since those machines only run HCC and C4CW.
In 1969 I took an oath to defend and protect the U S Constitution against all enemies, both foreign and Domestic. There was no expiration date.
|
||
|
|
Crystal Pellet
Veteran Cruncher Joined: May 21, 2008 Post Count: 1412 Status: Offline Project Badges:
|
Sek, thanks for your explanations.
Nice to know when starting with an empty task list. It looks like the calculation whether work should be requested is running once per minute. 22-Feb 09:28:48 work fetch resumed by user 22-Feb 09:29:43 Sending scheduler request: To fetch work. 22-Feb 09:29:43 Requesting new tasks for CPU |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
CP,
----------------------------------------Think indeed the buffer-up need is initially computed with a 1 minute interval, but incrementing if the network connection fails. Obviously if network is suspended per schedule, there wont be any attempts either. The upcoming 7 client will actually print a backoff time in the event log in addition to the counter that can be seen in the projects view. If you pull the internet cable and add <work_fetch_debug> to the <logflags> section of the cc_config.xml, then read it in, you'd see full detail how the backoff increases. If you'd be willing try this and let me know? It's one of these bits that can be added for the MMgrs. Have no any 6 client left to find out... all run the 7 Alpha [such sacrifice] ;D --//-- [Edit 1 times, last edit by Former Member at Feb 22, 2012 12:12:25 PM] |
||
|
|
|