| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: Website Support
Thread: Feedback on WCG upload-download process |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 3
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hello WCG
Attention: WCG-CAs/Techs In a twelve(12)-minute sync session with the WCG servers on 2010.05.29Sa.0452-through-0504.UTC, I was able to upload five(5) HPF2_v6.17 WUs while four(4) other HPF2_v6.17 WUs were left marked as readyToReport even as fourteen(14) new HPF2_v6.17 WUs were downloaded. Right at the start of the session, the download phase conducted simultaneousely with the upload phase. This is the first time I have seen this method. Unless I missed observing some of my sync sessions in the past, the method used was always: upload first, download next. If there was any mixing of the phases, it occurred at the tail end of the upload phase. The uploadFirst-downloadNext method appeared logical to me for it ties in well with the idea of the WCG server needing to know FIRST what WUs were done before making any decisions how to proceed NEXT in forwarding new WUs to crunchers. The mixing of upload and download phases may have somethng to do with why the said readyToReport WUs were not acknowledged in a straightforward and timely manner. In any case, in the subsequent one(1)-minute follow-up sync session on 2010.05.29Sa.0526-through-0527.UTC, I was able to have the said four(4) other HPF2_v6.17 WUs acknowledged...finally. For your feedback, please. ; |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
multi-thread/multi-task one task failed to complete while the other continued on. for the simple(ish) answer.
were there any errors reported in message log? may have been as simple as server was busy, and did not complete request in time, etc. correction after 1 min suggests this as maybe??? if is only a "one off" would say not a problem, if is a continuing problem, please post additional/again |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
andzgrid,
----------------------------------------The client scheduler decision process is a very complex piece of machinery. What you saw was just the circumstance causing it to happen simultaneous. Those who crunch off-line and only connect once a day like see this all the time... uploading result files and same time backfilling the buffer trying to always maintain at least it's stock at minimum setting i.e. at 1.0 days and the cache dropping below that level [including remaining time on work running], the scheduler will ask for X seconds of work. Because no task is exactly X seconds, you get one or more that are Y seconds, so the client can end up with a cache of 1.1-1.5 days, which will then stop any fetching until dropping below 1.0 days again. If interested, suggest to read the documentation at http://boinc.berkeley.edu/wiki/ and search for scheduler and scheduling and Feeder. Ready to Reports (RtR) are usually combined with work requests after the client has long uploaded the result files, the important parts [if it was allowed to do so]. RtR's go to the server side scheduler tracking what you have so it can decide how much more to give based on the client settings and pleadings for more.
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
|