| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: BOINC Agent Support
Thread: High Load Warning, scheduler request deferred |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 70
|
|
| Author |
|
|
BladeD
Ace Cruncher USA Joined: Nov 17, 2004 Post Count: 28976 Status: Offline Project Badges: 
|
11/7/2014 8:24:24 AM | World Community Grid | Message from server: We are currently experiencing high load and are temporarily deferring your scheduler request. Your client will automatically try again later. I now have 5 tasks waiting to report on 1 machine (M-09). 16 here. |
||
|
|
uplinger
Former World Community Grid Tech Joined: May 23, 2005 Post Count: 3952 Status: Offline Project Badges: 
|
We are looking into the issue, we have changed some settings but those have appeared to not look properly. It looks like it is starting to clear up now.
Thanks, -Uplinger |
||
|
|
BladeD
Ace Cruncher USA Joined: Nov 17, 2004 Post Count: 28976 Status: Offline Project Badges:
|
We are looking into the issue, we have changed some settings but those have appeared to not look properly. It looks like it is starting to clear up now. Thanks, -Uplinger Yep. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
who are you speaking from? i can't find him. i'm not able to see a spike in yesterday's runtime that would imply such a thing. I've searched on freedc for a user with high boinc points output yesterday and low total of boinc points. I've found user paina. I don't think that user is the culprit. They are returning results more than once a day. From the few times I checked, they have been returning results throughout the day. |
||
|
|
uplinger
Former World Community Grid Tech Joined: May 23, 2005 Post Count: 3952 Status: Offline Project Badges:
|
Ianbrown,
We have reached out to the member to try and help them configure their environment. But it appears that they are using a deep freeze or reimaging software on the computers once a day. This is forcing the agents to basically detatch and abandon all the results which could be over 2000 at one time. These are then attempted to be sent back out to Mac computers, but the problem with that is we don't have a lot of Mac computers also requesting work. Thus the waterfall effect. 1. Feeder is full of Mac only work units 2. Other computers (Windows, Linux, Android) can't download work and are told to back off. 3. This causes the scheduler to start getting over worked especially when a few slots open. This is basically a traffic jam of sorts for the rest of the grid. 4. When scheduler is over worked, it tells computers to back off due to high load. So as you can see a waterfall effect. The problem is that during a majorit of the day, things are doing well. And making changes doesn't show any difference. This is happening during the 3-4 AM in the morning my time, which is usually when I'm sleeping. Thus I can only make some changes and hope they work over night. I have implemented a new change to the scheduler and feeder to put slots only based on the percentage of usage on the overall grid. This means that if windows uses 80% of the grid, then 80% of the slots should be available to Windows only. Thus in theory preventing Mac from filling up all 2700 slots. Thanks, -Uplinger |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Was there not a rule set by your former colleague to initially give little work, but after 3 hours open the gauges? Extend that rule to 24 hours, since they work on a daily cycle. In effect, if a device persists and keeps returning work then it's ok to buffer more, else no way josé.
Surely the adverse is those complaining of not getting enough, like me as just experienced in a dual boot environment, but a reputation for an extablished device returning for new work is not expected to be impacted. Yes/no  |
||
|
|
uplinger
Former World Community Grid Tech Joined: May 23, 2005 Post Count: 3952 Status: Offline Project Badges:
|
The problem is, the machines reattach as the same machine, however the results were marked as client detached, so the resend mechanism that we have in place, does not allow for those results to be sent back to the same host when it attaches to World Community Grid again. I am looking to see if I can change the scheduler to include that case as a reason for finding lost results.
I'm not sure why you're not getting enough work, it should be acting as normal as most of the results in the feeder are marked for any OS at this time. Thanks, -Uplinger |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
11/7/2014 2:18:58 PM | World Community Grid | Scheduler request failed: Couldn't resolve host name 11/7/2014 2:19:27 PM | | Project communication failed: attempting access to reference site 11/7/2014 2:19:33 PM | | Internet access OK - project servers may be temporarily down. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
knowing the exact issue is half the fix, so hope is on
|
||
|
|
LouisonF
Cruncher Joined: May 8, 2011 Post Count: 4 Status: Offline Project Badges: 
|
Hi,
----------------------------------------I haven't any task from WCG for approximately 2 days. I haven't read all the subject but as i can understand (i'm not sure of what i understand because i'm not american or english so i don't understand everythings :) ) i'm not alone. I got 0 error or anything like that but just i can spam the actualise button but i never got any task from WCG. So please if you got a solution or something like that ( i had read u are trying to solve this but i'm not sure i got the exactly same problem) Thanks you PS : Sorry for my bad english, hope you will understand me ;) [Edit 3 times, last edit by loulou96 at Nov 7, 2014 10:53:41 PM] |
||
|
|
|