| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Official Messages
Forum: News
Thread: Regarding ARP1 and MCM1 download issues since ARP1's launch on Monday Nov 4th, 2024 |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 159
|
|
| Author |
|
|
nairb
Cruncher Joined: Jun 30, 2020 Post Count: 3 Status: Offline |
no improvement yet. Maybe better later
|
||
|
|
Seoulpowergrid
Veteran Cruncher Joined: Apr 12, 2013 Post Count: 817 Status: Offline Project Badges: 
|
I am also getting more than the last few days. Still have WUs from another project to keep my CPUs fully busy, but I am getting more from WCG.
---------------------------------------- |
||
|
|
ludarp
Cruncher Joined: Nov 5, 2011 Post Count: 4 Status: Offline Project Badges: 
|
(Merging a (now deleted) thread here.)
----------------------------------------I thought it would be worth quantifying the impact of APR / the isuse upon MCM, based on my reading of the projects' stats pages. MCM before November performed roughy 270 Years of computation per day, but this dropped to ~80 Years/day since. Meanwhile, ARP computation has only risen to 7 Years/day. So I assume most of the lost computation of MCM is due to stalled BOINC clients (with CPUs idle or shifted to WCG projects) since BOINC will not download more work-tasks whilst even a single download-task is pending (and usually there are many). (And personally, I am still "babysitting" with the Retry option on 2 PCs to avoid this stalling, which isn't great for work productivity.) So I would say the idea to "decrease the upper limit on workunits to produce" (quoting the first post in this thread) - for the APR workunits - sounds good, at least until the bigger scaling issues get worked through. [Edit 1 times, last edit by ludarp at Nov 11, 2024 2:30:18 PM] |
||
|
|
Boca Raton Community HS
Advanced Cruncher Joined: Aug 27, 2021 Post Count: 125 Status: Offline Project Badges: 
|
I'd be happy to but how am I reducing the efficiency of their network if all the work units my boxes download are being computed successfully and submitted successfully? And not expiring, evidence is no 'too late' work units That is great if all of the files are downloading and they are processing fully. It is so odd to me that some users have no issues downloading files and others (such as us) are having a REALLY difficult time. |
||
|
|
Link64
Advanced Cruncher Joined: Feb 19, 2021 Post Count: 129 Status: Offline Project Badges: 
|
I hope I am not contributing to the problem with my farm set to 10 days buffer Considering the 6 days deadline for MCM you might a bit as some WUs are likely not started before deadline and need to be resend to someone else. Buffer should be always significantly below the deadline.Thank you! But I have on all machines set to 10 days and 'up to 10 days more' however I just looked on my Results page ( https://www.worldcommunitygrid.org/contribution/results?validationStatusIds=5 ) sorted 'too late' , 0 results so I seem to not be wasting their bandwidth. No errors (if that's the errors page here, all other projects have state=6 for that) means, that you start all tasks before the deadline, not that you finish and report them to the server before the deadline so that replacement tasks don't need to be created and send out to another computer. I see sometimes in my task list such "aborted by server" task returned just a bit too late. Tasks returned after deadline, but before the replacement task is returned validate and do not appear as an error in the list.  [Edit 1 times, last edit by Link64 at Nov 11, 2024 5:48:56 PM] |
||
|
|
danwat1234
Cruncher Joined: Apr 18, 2020 Post Count: 39 Status: Offline Project Badges:
|
I'd be happy to but how am I reducing the efficiency of their network if all the work units my boxes download are being computed successfully and submitted successfully? And not expiring, evidence is no 'too late' work units That is great if all of the files are downloading and they are processing fully. It is so odd to me that some users have no issues downloading files and others (such as us) are having a REALLY difficult time. Agreed, seems to favor certain IPs maybe distance? I don't know why every time WCG has trouble the servers like me. Best i can say is have another project on standby keep on crunching there's plenty of biological work unit love to go around Rosetta Sidock Denis or our neighbors; Folding 0x08 CPU. let's change the world! In my Results, filters being; error, no reply, validation error, abandoned, invalid, workunit error, too late; Total 1883. Makeup; detached 659 error 1222 log shows nothing on any i have checked +1 server aborted + 1 unaccounted for. Goes back to November 5th so 6 days 1883 results.. According to my overview the day chart. Completed 110160 WUs from November 5th to today so 1.7% are questionable so seems as though my farm is optimal. [Edit 1 times, last edit by danwat1234 at Nov 11, 2024 6:27:01 PM] |
||
|
|
ericinboston
Senior Cruncher Joined: Jan 12, 2010 Post Count: 258 Status: Offline Project Badges: 
|
Long time WCGer here...I've got about 15 machines, mostly Macs, that have been only returning 1/10th to 1/5th the WUs they normally crunch over the past ~week. I only process MCM. Currently I do not have the time to connect to all the different Macs (only have 1 keyboard and 1 monitor) to write down details of exactly what is going on with BOINC so right now I cannot tell you if there are errors in the client, or WUs that are stuck at 13%, or some other info.
----------------------------------------Can someone give us a laymens' answer on what we should do (if anything)? Such as, should I reboot the Macs? Or kill all the tasks and ask for more? Or detach from MCM and reconnect? Or change some preference in the client? On a side note, it sure would be nice if WCG could post something on this (very important) thread about what the overall fix is on their end, when it will be fixed, etc.  [Edit 1 times, last edit by ericinboston at Nov 11, 2024 7:09:52 PM] |
||
|
|
Mike.Gibson
Ace Cruncher England Joined: Aug 23, 2007 Post Count: 12348 Status: Offline Project Badges: 
|
Downloading seems to have improved almost to free-flowing now the initial rush is about over.
Mike |
||
|
|
Falconet
Master Cruncher Portugal Joined: Mar 9, 2009 Post Count: 3295 Status: Offline Project Badges: 
|
I did download a good bunch of tasks just now.
----------------------------------------AMD Ryzen 5 1600AF 6C/12T 3.2 GHz - 85W AMD Ryzen 5 2500U 4C/8T 2.0 GHz - 28W AMD Ryzen 7 7730U 8C/16T 3.0 GHz |
||
|
|
TPCBF
Master Cruncher USA Joined: Jan 2, 2011 Post Count: 1948 Status: Offline Project Badges: 
|
Can someone give us a laymens' answer on what we should do (if anything)? Such as, should I reboot the Macs? Or kill all the tasks and ask for more? Or detach from MCM and reconnect? Or change some preference in the client? I would recommend NOT to change any preferences. This will just make things worse (for everyone, though some folks don't seem to grasp that issue).Anyway, as for MCM1, just manually retry the transfers until you have a working number of WUs, then let the rest settle on its own. Things seem to run a bit more smoothly today than they did for the end of the weekend, though I wouldn't call it "free flowing" as Mike Gibson put it... Rebooting,Killing tasks or detaching/reattaching is just making the situation worse for YOU. By the way, that 13% of seemingly stuck downloads is due to those WUs just downloading a 107 bytes long error message, not any part of the 912/915 bytes sized actual WU download file for MCM. On other projects (currently only ARP1) that percentage will vary depending on the size of the WU file and the ratio of that size compared to the 107 bytes error message... Ralf  |
||
|
|
|