| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: BOINC Agent Support
Thread: One Computer Will Not Report Automatically |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 11
|
|
| Author |
|
|
jmcgaw
Advanced Cruncher US Joined: Feb 2, 2007 Post Count: 54 Status: Offline Project Badges: 
|
I have five computers running WCG but one has a sudden aversion to reporting results and downloading new units when it needs to. I don't see a network problem since, when I 'update' WCG from the panel, it does what it should. Subject machine is an i7 running W7-64 and version 7.0.25(x64). So far as I know there were no changes to any settings before the problem cropped up although I'll admit to a load of them since I noticed it. This machine runs my default profile with the exception that it is limited to 75% of processors because running all eight makes the fans just a bit too noisy for the office.
Any clues on how to troubleshoot? |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Suggest you read up on the 7.0.xx client way of reporting and work fetching [there's a draft FAQ in the support forums]. Depending on your settings, inherited from the 6.xx client, you'll need to get used to new ways [more efficient to the servers]. Work wont be fetched until the Minimum buffer is reached [formerly known as "Connect about every.."]. This work fetching is combined with the Ready to Report clearing. If your connect was at 0.00 days, then it means your client wont go and fetch work until the minimum buffer has reached 0.00 days in cached work.
------------------------------------------//-- [Edit 1 times, last edit by Former Member at Apr 26, 2012 8:50:11 PM] |
||
|
|
jmcgaw
Advanced Cruncher US Joined: Feb 2, 2007 Post Count: 54 Status: Offline Project Badges:
|
Thanks, I'll check that out. I'll also have to check since at least one other nearly-identical computer (processor, ram, disk, os, client) has never shown this behavior. Of course that machine lives in the corner of the basement so it could be that I just don't notice what it is doing as long as it is doing something.
|
||
|
|
mikey
Veteran Cruncher Joined: May 10, 2009 Post Count: 824 Status: Offline Project Badges:
|
Thanks, I'll check that out. I'll also have to check since at least one other nearly-identical computer (processor, ram, disk, os, client) has never shown this behavior. Of course that machine lives in the corner of the basement so it could be that I just don't notice what it is doing as long as it is doing something. You could always make a small cc_config.xml file and put a line in it to report results immediately upon completion. This is not a 'normal' thing to leave in there but it can be used it needed. Here is a small cc_config.xml file: <cc_config> <options> <report_results_immediately>1</report_results_immediately> </options> </cc_config> Copy and paste the above lines into NOTEPAD and save it as a txt type file, DO NOT USE A WORD PROCESSING TYPE PROGRAM as it adds stuff that makes the file not run!! Put the file in your Boinc directory and restart Boinc and you should be good to go. Here is where you can find more stuff about the cc_config.xml file: https://boinc.berkeley.edu/w/?title=Client_co...;diff=3292&oldid=prev   |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Thanks - this works well. Mine was only reporting every 7 hours and I am running in a 1 week points challenge this week.
|
||
|
|
mikey
Veteran Cruncher Joined: May 10, 2009 Post Count: 824 Status: Offline Project Badges:
|
I have five computers running WCG but one has a sudden aversion to reporting results and downloading new units when it needs to. I don't see a network problem since, when I 'update' WCG from the panel, it does what it should. Subject machine is an i7 running W7-64 and version 7.0.25(x64). So far as I know there were no changes to any settings before the problem cropped up although I'll admit to a load of them since I noticed it. This machine runs my default profile with the exception that it is limited to 75% of processors because running all eight makes the fans just a bit too noisy for the office. Any clues on how to troubleshoot? As SekeRob said CHECK YOU CACHE SETTINGS in the Boinc Manager, they have changed how they work in version 7! The left side is now the MINIMUM time to connect, so if you 4.25 in there it will ONLY connect once every 4.25 days! Change the left side number to a small one, I use 0.75, and then put something like 0.50 in the right side, this gives me about 1.25 days of work. This works for people with always on connections, yours may need to be changed! I keep a small cache, if you like a bigger cache adjust the right side number to a bigger one, start SMALL and work up. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
To augment in knowledge [from experience and testing] running a MinB/MaxAB both set to zero [0.00], will actually do report clearing immediately, at least when there are so many jobs on the client and running as there are active threads in a client. Work will actually be obtained [when work fetch is not suspended] 5 minutes before any task finishes, so there's a seamless operation, not like CEP2 needing 7-10 minutes to upload before a next task is fetched]. This is how I run the Ubuntu 12.04 install on USB 3.0 memstick. It's a bonus to those running on edge [least cached work], fastest possible turn around time.
----------------------------------------Further, [moi] requested a scheduling enhancement which was implemented in 7.0.23 or .24 [which like .25/.26 have other serious scheduling bugs, so that .27 is already in the works which is without a doubt going to be followed shortly on with a .28 test client], that will clear any RtR 30 minutes prior to a scheduled network out [it truly works on my 7.0.24 & .26 hosts]. This then got an addition to do same when computing is suspended per schedule. The RtR clearing is continued until the actual network or computing suspend. You work out a new trick how to force with this an EOD [UTC] clearing, which of course is no guarantee that the validators will get to it. They're very busy prepping for the statistics run right around that time. --//-- [Edit 1 times, last edit by Former Member at May 1, 2012 1:20:23 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Those that jumped on the 7.0.25 bandwagon and any others that have any other 7.0.xx may want to fetch 7.0.27, just out. Whole boat load of fixes and polishes in the Change Log. My focus will be to see if it will NOT fetch work which I had again this morning with a 7.0.26... the client has to be unloaded and loaded again. 2 of 8 cores were idling due this, when the design is to fetch work no later than 5 minutes before any active core is threatened to go idle.
Get it here http://boinc.berkeley.edu/dev/forum_thread.php?id=6698&nowrap=true#43938 and be so kind to report at the developers forum if you have issues (or an the alpha mail list if a subscribed tester). --//-- P.S. Intermediate reports on the dev forum suggest the issue is fixed. Don't know how they know in a few hours, when it takes me crunchers days to stumble on these bugs. |
||
|
|
Johnny Cool
Ace Cruncher USA Joined: Jul 28, 2005 Post Count: 8621 Status: Offline Project Badges: 
|
Thanks, I'll check that out. I'll also have to check since at least one other nearly-identical computer (processor, ram, disk, os, client) has never shown this behavior. Of course that machine lives in the corner of the basement so it could be that I just don't notice what it is doing as long as it is doing something. You could always make a small cc_config.xml file and put a line in it to report results immediately upon completion. This is not a 'normal' thing to leave in there but it can be used it needed. Here is a small cc_config.xml file: <cc_config> <options> <report_results_immediately>1</report_results_immediately> </options> </cc_config> Copy and paste the above lines into NOTEPAD and save it as a txt type file, DO NOT USE A WORD PROCESSING TYPE PROGRAM as it adds stuff that makes the file not run!! Put the file in your Boinc directory and restart Boinc and you should be good to go. Here is where you can find more stuff about the cc_config.xml file: https://boinc.berkeley.edu/w/?title=Client_co...;diff=3292&oldid=prev I too have got a similar problem. Bad enough that I almost want to stop crunching (I have been crunching for over 7 years now). My one PC running 24x7 and yet a many results were being help back for some time now. I'm sure this script would greatly help me as well. What I don't understand is that you say 'save as text", yet you did not post whether you later rename this as cc_config.xml and then copy it to your Boinc Data folder. I hope you can help me. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
There is *no* reason why Ready to Report do not report by themselves within 24 hours of the oldest RtR listed having completed. High caching with variability of run time, causing say a 3 day cache suddenly becoming 6 days, means the client does not fetch work for 3 days, or however long it takes till the cache sinks below the 3 day setting. Even then, if a net connection is available, the client will seek to report these Ready to Reports no later than 24 hours from done and result files uploaded.
----------------------------------------The said instruction to insert that line is *not* appreciated by WCG [no project in fact]. It means that if someone is completing 2000 results per day on a host, the client will connect 2000 times per day to the servers, and greatly contribute to overall server load impacting everything else... tardi feeders, tardi assimilators, tardi validators, so thank you for not setting that config **. ** The option has a very special application object... to help those that re-image their computer every day, who then otherwise would loose any completed tasks that have not reported. [Edit 1 times, last edit by Former Member at Jan 20, 2013 7:00:15 PM] |
||
|
|
|