| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: Website Support
Thread: Tasks won't finish in time sent in error |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 16
|
|
| Author |
|
|
Mike.Gibson
Ace Cruncher England Joined: Aug 23, 2007 Post Count: 12360 Status: Offline Project Badges: 
|
I have my cache set to 10 days and am only requesting Zika & Aids2. I have app_config set to 2 units of Aids2 plus 6 zika (only 8 threads on my system). In order to get 24 hours of Aids2 (8 WUs), I have had to reduce my Zika cache to 48 WUs. This is only 24 hours of Zika crunching.
The message that I have been receiving is: 25/05/2019 13:03:25 | World Community Grid | Tasks won't finish in time: BOINC runs 97.0% of the time; computation is enabled 100.0% of that I realise that Aids2 is restricted to a 24-hour turnaround but Zika has a 10 day limit. Any suggestions? Incidentally, the spell-checker on this forum does not recognise the word Zika. The project names should be added to the spell-checker! Mike |
||
|
|
Sgt.Joe
Ace Cruncher USA Joined: Jul 4, 2006 Post Count: 7662 Status: Offline Project Badges: 
|
The easiest thing is to just drop your cache setting to something like 8 days and see if that cures the problem. I run with 1 day cache of my Zika only machines and the supply for Zika has been rock steady. I would keep dropping my cache until the problem message disappears.
----------------------------------------Cheers
Sgt. Joe
*Minnesota Crunchers* |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Mike,
I'm not sure I understand the whole picture here, but one thing I do know is that BOINC doesn't handle app_config as well as it might. I don't think BOINC is taking that restriction into account when it's doing its scheduling calculations. But if you're running a sub-project with a 24-hour deadline and not letting your machine run all it can, then things are soon going to clog up. And anyway, what's the point of having a cache of more than a day? You might be able to improve the situation if you set a limit on the number of AIDS units that the host can send you, by using the limits that are now available in the profile (editable via the Device Manager page). That way you should be able to ensure that the cache of AIDS units don't sit for more time than they can be processed within, even if you have a larger cache of Zika units. Also, unless there's something else I don't understand, any spell checking you're seeing in this forum is being done by your browser, not by the forum. Just add any unknown words to your browser's dictionary by using mouse button 2 on the wavy red line (or whatever you see to tell you something is spelled wrongly) and selecting "add to dictionary". |
||
|
|
Mike.Gibson
Ace Cruncher England Joined: Aug 23, 2007 Post Count: 12360 Status: Offline Project Badges:
|
Thanks for the replies. That has cured the spell-check problem, but not the cache problem.
I have 8 threads, split 2 Aids2/6 Zika by app_config, with unit times of 5.5 hours and 2.9 hours respectively. This means that the maximum Aids2 cache of 8 units is completed in 22 hours (inside the 24 hour limit). However, in order to get the 8 Aids2 units I have to drop my Zika profile to 48 units which take only 23 hours whereas the time limit is 10 days. On this configuration, there is no way that the tasks won't finish in time, but I still get the message regularly. With crunching averaging 97% of the time, I should be able to hold 8 Aids2 and 480 Zika units, but anything more than 48 Zika would cut my Aids2 cache. 24 hours of each is normally enough as I am on-line 24/7 but my connection is not always reliable, so like to hold a larger cache. I am just trying to draw attention to the fact that the message comes up when it is not actually correct as my cache will finish within the time allowed. Also that the Aids2 causes the Zika cache to be restricted unnecessarily. I will be suspending Aids2 crunching in a couple of months when I get to 5 years and will then crunch Zika to the same level and MCM to 20 years. I like to crunch 2 at a time in case of lulls in production. Mike |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Mike,
If you limit AIDS to 8 in your device profile, it shouldn't matter if you leave Zika as unlimited. Similarly, if your cache is larger than the return time limit for AIDS, they should all run as high priority and so another should start as soon as one finishes. Having said all that, I did try to play with something along these lines and it didn't work well, but that was before the profile limit was available. I THINK that it should now work. If not, my gut says that there may not be too much that can be done without a major re-think of BOINC because, if my guess is correct, the client does not send info about app_config restrictions back to the host. But I might be talking through what passes for my hat ... |
||
|
|
AgrFan
Senior Cruncher USA Joined: Apr 17, 2008 Post Count: 376 Status: Offline Project Badges: 
|
I agree with Sgt. Joe. No need for a 10 day cache. Drop it to 2-3 days and the message should go away.
----------------------------------------I have a 8 thread machine set to 4 FAH2 and 4 MIP units in app_config. Device profile is set to unlimited for FAH2 and 8 for MIP. I always have 16 total units downloaded ... 4 running and 4 queued of each project. Works perfectly ... equal daily production for each project. [Edit 1 times, last edit by AgrFan at May 26, 2019 3:24:43 AM] |
||
|
|
Mike.Gibson
Ace Cruncher England Joined: Aug 23, 2007 Post Count: 12360 Status: Offline Project Badges:
|
Thanks folks. I have set my cache to 3 days and both projects to unlimited. 505 Zika units have arrived but I will wait until it stabilises before reporting back. The current message says that my computer has reached its limit. app_config is still set to 2 Aids2 and 8 Zika but is operating as if 2/6 as the Aids2 units are closer to their time limit.
I reckon that I now have 10 days cache, instead of the 3 set in device profile, and suspect that my Aids2 units will not renew as they complete. Mike Mike |
||
|
|
AgrFan
Senior Cruncher USA Joined: Apr 17, 2008 Post Count: 376 Status: Offline Project Badges:
|
Mike,
----------------------------------------Change your device profile setting for ZIKA to 64. That will allow 1.5 days of ZIKA work in the queue. With the cache set to 2-3 days, that should keep FAH2 units running as they complete. Let us know if there's improvement once the machine stabilizes. [Edit 1 times, last edit by AgrFan at May 26, 2019 1:20:14 PM] |
||
|
|
Mike.Gibson
Ace Cruncher England Joined: Aug 23, 2007 Post Count: 12360 Status: Offline Project Badges:
|
Update
While watching F1, I was not monitoring WCG. Currently I have 555 Zika units In Progress and they are being replaced as they are finished. My Aids2 units are not being replaced on completion so are down to 5 units. So, I have now got 67 days Zika cache and 14.5 hours Aids2 cache. That is a total cache of 8.5 days, based on 8 threads on my machine. Despite reducing my cache target to 3 days, it has increased by 7.5. That is obviously not working correctly. I will now reduce my device profile to 64 Zika and abort 490 units and wait to see what effect that will have on my Aids2 cache. Mike |
||
|
|
AgrFan
Senior Cruncher USA Joined: Apr 17, 2008 Post Count: 376 Status: Offline Project Badges:
|
Despite reducing my cache target to 3 days, it has increased by 7.5. That is obviously not working correctly. The cache may be still set to 10 days locally which is overriding the device profile setting. Make sure the cache is set to 3 days in the device profile and click 'Use Web Prefs' under 'Options - Computing preferences...' on the machine to reset the local preferences. [Edit 3 times, last edit by AgrFan at May 26, 2019 10:26:22 PM] |
||
|
|
|