| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Active Research
Forum: Africa Rainfall Project
Thread: Work Available |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 3596
|
|
| Author |
|
|
Unixchick
Veteran Cruncher Joined: Apr 16, 2020 Post Count: 1312 Status: Offline Project Badges: 
|
I wish people would set reasonable queues. It is probably pointless to post this here, as you guys know your queue and keep an eye on it. It puts a load on the server for it to have to resend the WUs. The server is already having download issues, without having to do resends. ARGGG! Thanks for letting me vent as my list of pending grows.
|
||
|
|
TLD
Veteran Cruncher USA Joined: Jul 22, 2005 Post Count: 860 Status: Offline Project Badges: 
|
I think part of the reason that folks run large queues is that ARP WUs are not always available and that is all they run.
----------------------------------------I have computers that can finish a ARP WU in less than a day, those computers are set to download 2 WUs. I also have computers that take 3 days to run a ARP WU and those computers are set to download 1 ARP WU. I also run all projects so my computers are always busy. and with a 2 day que I only have problems when the download problem raises it's ugly head.  |
||
|
|
paulch2
Cruncher Joined: Aug 6, 2020 Post Count: 25 Status: Offline Project Badges:
|
Since a task is listed as in-progress as soon as the download starts, I wonder how many are "late" before they fully download. But yes, the people with deep queues are annoying. I've got 27 pages of tasks pending validation for MCM, and for most, the other machine times out before it's finally handed off to one that can complete it. Not quite so bad for ARP tasks, but it's still noticeable. Would be nice if they updated the software to pair up groups of machines which had similar elapsed times, especially for serial tasks like ARP.
|
||
|
|
Unixchick
Veteran Cruncher Joined: Apr 16, 2020 Post Count: 1312 Status: Offline Project Badges:
|
I'm starting to get ARP resends now. Not sure if original ARPs are still going out.
I'm only downloading 1 WU at a time (queue of 3 for 2 active cpus), so I can image with multiple WUs downloading at a time you get bits of each one, but takes longer to download 1 full one. It would be good if it could prioritize downloading bits of one WU at a time. I can see in my log that it took 30 minutes to fully download the requested WU last night, without intervention. I'll retry manually to get the bits of WU if I'm at the computer at lunch time or dinner time, but otherwise just having one extra hasn't given me idle time unless the hopper is truly empty. |
||
|
|
Mike.Gibson
Ace Cruncher England Joined: Aug 23, 2007 Post Count: 12594 Status: Offline Project Badges: 
|
I only seem to getting initial normals but my last was 10 hours before your post (at 08:16 GMT (UTC).
Mike |
||
|
|
mikey
Veteran Cruncher Joined: May 10, 2009 Post Count: 826 Status: Offline Project Badges: 
|
Since a task is listed as in-progress as soon as the download starts, I wonder how many are "late" before they fully download. But yes, the people with deep queues are annoying. I've got 27 pages of tasks pending validation for MCM, and for most, the other machine times out before it's finally handed off to one that can complete it. Not quite so bad for ARP tasks, but it's still noticeable. Would be nice if they updated the software to pair up groups of machines which had similar elapsed times, especially for serial tasks like ARP. Alot of that problem is new people still using the default 10 day cache in the Boinc Manager, I've always thought that was a mistake to keep it like that for this time of alot of people having cable and not needing a huge cache of tasks anymore   |
||
|
|
mikey
Veteran Cruncher Joined: May 10, 2009 Post Count: 826 Status: Offline Project Badges:
|
I'm starting to get ARP resends now. Not sure if original ARPs are still going out. I'm only downloading 1 WU at a time (queue of 3 for 2 active cpus), so I can image with multiple WUs downloading at a time you get bits of each one, but takes longer to download 1 full one. It would be good if it could prioritize downloading bits of one WU at a time. I can see in my log that it took 30 minutes to fully download the requested WU last night, without intervention. I'll retry manually to get the bits of WU if I'm at the computer at lunch time or dinner time, but otherwise just having one extra hasn't given me idle time unless the hopper is truly empty. When I click retry on a download that is delayed it goes to the top of the list to be downloaded next, I can see this by watching the name of the workunit as it moves up to the retry list where I click to restart the download and boom it goes to the top of the retry list to be downloaded immediately again and again until it finally goes thru. I have 17 pc's running but not all of them are running WCG, obviously, but the ones that are are running either the ARP tasks OR the gpu tasks but BOTH have the constant retries in their downloads. One problem with the gpu tasks is they are sooo fast that you can be downloading and uploading tasks at the same time and then of course the list gets longer as you returned a task so you need more. I see similar things for the ARP tasks where i click to retry the downloads and boom and whole list of new tasks to be downloaded shows up and of course needs to be retired a dozen times to get them to finally come thru. Each of my pc's has a 1.5 day cache setup which of course is different for each pc based on the number of cpu's that are in them. |
||
|
|
mikey
Veteran Cruncher Joined: May 10, 2009 Post Count: 826 Status: Offline Project Badges:
|
I only seem to getting initial normals but my last was 10 hours before your post (at 08:16 GMT (UTC). Mike I'm getting _0, _1, _2 and _3 tasks for ARP tasks but only _1 and _2 tasks for gpu tasks |
||
|
|
Mike.Gibson
Ace Cruncher England Joined: Aug 23, 2007 Post Count: 12594 Status: Offline Project Badges:
|
Resends within the first half of the initial period don't really count as resends as they still get the the initial period. Later resends get half the initial period.
Mike |
||
|
|
catchercradle
Senior Cruncher England Joined: Jan 16, 2009 Post Count: 171 Status: Offline Project Badges: 
|
I'm only downloading 1 WU at a time (queue of 3 for 2 active cpus), so I can image with multiple WUs downloading at a time you get bits of each one, but takes longer to download 1 full one. It would be good if it could prioritize downloading bits of one WU at a time. I raised this on the BOINC boards recently. It is a requested feature over at git-hub where the code is worked on but it has been for quite a long time with no one leaping up to own and work on it. This is the link to my last post about it there. |
||
|
|
|