| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: Help Cure Muscular Dystrophy - Phase 2 Forum
Thread: Do not think it helps to stock up large caches... |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 39
|
|
| Author |
|
|
Mysteron347
Senior Cruncher Australia Joined: Apr 28, 2007 Post Count: 179 Status: Offline Project Badges: 
|
Then, of course, a large cache does help when, say, the host has a file-system problem....
|
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
The definition of "Large" is to me more than 2 days. As this proofed, the laptop only ran out of WCG work on 1 core with the 0.9 day buffer setting as it happened that the HFCC jobs ran longer than projected. Now I've increased it to 1.2 days which I guess will be save for the next few years.
----------------------------------------
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
Hypernova
Master Cruncher Audaces Fortuna Juvat ! Vaud - Switzerland Joined: Dec 16, 2008 Post Count: 1908 Status: Offline Project Badges: 
|
I cache 3 days work. That should cover 99.999999% of the outages that WCG I agree with Movieman on that, and indeed it is the standard value for me too. But there is another reason to size the cache to a certain value. It happened for unknown reasons that the WLan link that I have on each of my machines does go down. Certain devices are able to reconnect others not. As I travel often, for three days at least at a time, I cannot attend to correct problems. To avoid my machines to stop crunching and stay running idle but nevertheless consume power during my absence, I have set the cache to 3 days which does cover the period. If I would be away for a few weeks ex. during holidays, I may occasionally put it to 10 days and then switch back.  [Edit 2 times, last edit by Hypernova at Oct 16, 2010 9:57:30 AM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Then, of course, a large cache does help when, say, the host has a file-system problem.... and that happens about once every 3 years |
||
|
|
sk..
Master Cruncher http://s17.rimg.info/ccb5d62bd3e856cc0d1df9b0ee2f7f6a.gif Joined: Mar 22, 2007 Post Count: 2324 Status: Offline Project Badges: 
|
I had a similar problem trying to use a USB Wireless dongle on one system with an entry level GPU. I had to keep a 0.01 level cache for the GPU, but wanted a higher cache for the CPU as it regularly lost its connection, a real pain. Had to select No new GPU tasks, then up the cache to about 5days, download CPU tasks and reduce the cache back to 0.01days. I often had the problem of Boinc asking for GPU tasks only. It usually had to raise the cache to about 3.5+ days before Boinc would even start asking for CPU tasks.
Is there a command to report tasks on exiting Boinc? Would be a useful default. |
||
|
|
Mysteron347
Senior Cruncher Australia Joined: Apr 28, 2007 Post Count: 179 Status: Offline Project Badges:
|
and that happens about once every 3 years OK - I'll plead guilty to gratuitous exploitation of topicality. Nevertheless, since the last incident was what - 18 months ago - ? I'd suggest you're underestimating the problem by about 50% Problems can happen both ends, of course. I recently had a fan failure in a hub. The consequences were that whereas the visible operation of the machines involved (old slow print-servers) appeared unhindered, the system became completely BOINCstipated - all BOINC communications appeared blocked. Since I only check these machines weekly at best, it was only by chance I found there was a problem. Strangely, Firefox had no apparent problem accessing several sites - only BOINC appeared to be affected. The solution was to spend a merry few hours at the dead of night in mid-winter re-laying cables to bypass the hub and communicate directly with the modem. This flushed the system and downloaded new work - but since then, I have noticed that the modem seems to have a frog in its throat, so having replaced the fan, I'll return the system to its original configuration. Naturally, this frog has caused the same symptoms - and this time it was coincident with the other end's being unavailable. Murphy rules. |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
The original discussion stemmed from the overlong Pending Validation times on the HCMD2 project specifically. Whatever volunteer caching change or WCG change, the topic has in that become largely redundant for this quorum 2 work distribution as when I've been crunching this research in the mix semi constantly, the last periods PV jail for this science has largely been broken down, rarely going over 2 days.
----------------------------------------Cache, but keep it in check is what we're asking. Hypernova, there are keep-alive routines for WIFI... it could be in the power saving realm. I've few days ago put a Linux crontab job in to restart network-manager periodically and it is: Stable! The DHCP IP assignment list says the lP lease is 14 days minus 3 days and counting before expiring.
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
Allen008
Senior Cruncher USA Joined: Sep 22, 2009 Post Count: 244 Status: Offline Project Badges: 
|
I have my cache set to 0.1 day or something like that. It works for me. Once in a while a server problem will occur, and I'll pick up a few WU for other projects, and that is OK with me. As long as my Macintosh has something to crunch, I'm happy. Under "normal" conditions, I'll choose the projects, and short detours to other projects don't upset me.
---------------------------------------- |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
WCG's 0.3 days is pretty good, in case the internet itself goes down at your end ;-)
----------------------------------------
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
|