Advanced search

Message boards : Server and website : Problem resolved: new WUs stuck in "downloading" status

Author Message
joeybuddy96
Send message
Joined: 1 Apr 20
Posts: 3
Credit: 33,432,768
RAC: 430,306
Level
Val
Scientific publications
wat
Message 54975 - Posted: 27 May 2020 | 2:56:40 UTC

I came back after about 24 hours of not using my computer to discover that GPUGRID tasks were stuck. There were two of them in the tasks queue that were stuck on "downloading." There were several m'dad Toni files that were stuck on "downloading" in the transfers tab. There were no GPUGRID tasks running at the time.

I tried hitting "update" on the GPUGRID project in the projects tab. That did nothing. I hit "retry now" for each of the stuck files in the transfer tab. It was going to take like three or more minutes, but I didn't feel like waiting.

I reset the GPUGRID project in the projects tab. Some GPUGRID tasks appeared in the tasks tab, but they were still stuck in "downloading." I went to the scheduler tab and some new files from the whole GPUGRID project were downloading, but then it got stuck on some of the files. Their download progression wasn't moving, even though the files are relatively small and I'd already checked the GPUGRID server status earlier so I didn't think that was the problem.

What ended up fixing the problem was suspending network activity and then resuming it in the BOINC client. That was after I had reset it. Once the initial project files had downloaded, GPUGRID tasks started downloading and running normally again.

It's weird that this happened after it had already been working on the new batch of tasks that were released a few days ago, unless there was another round of project updates I missed. I think I had reset it back on the first day the tasks were available, but I could be wrong.
____________

Aurum
Avatar
Send message
Joined: 12 Jul 17
Posts: 252
Credit: 9,791,563,847
RAC: 2,648,925
Level
Tyr
Scientific publications
wat
Message 54985 - Posted: 27 May 2020 | 12:15:27 UTC
Last modified: 27 May 2020 | 12:25:56 UTC

Mostly worked for me. In BoincTasks it's Extra/Allow Network Communication then select the computers with stuck WUs and click Never followed by Always. Then Retry All. May be a delayed reaction.

One recalcitrant WU is still hung at Download pending.

Richard Haselgrove
Send message
Joined: 11 Jul 09
Posts: 1003
Credit: 2,537,540,285
RAC: 3,303,364
Level
Phe
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 54986 - Posted: 27 May 2020 | 12:24:21 UTC - in response to Message 54985.

This didn't work for me. WUs are still hung at Download pending.

Toni has posted that you can change the download url from gpugrid.org to gpugrid.net, if you don't want to abandon https.

Aurum
Avatar
Send message
Joined: 12 Jul 17
Posts: 252
Credit: 9,791,563,847
RAC: 2,648,925
Level
Tyr
Scientific publications
wat
Message 54988 - Posted: 27 May 2020 | 12:29:10 UTC - in response to Message 54986.

This didn't work for me. WUs are still hung at Download pending.

Toni has posted that you can change the download url from gpugrid.org to gpugrid.net, if you don't want to abandon https.
The problem persists for both http and https. How does https solve it?

mmonnin
Send message
Joined: 2 Jul 16
Posts: 284
Credit: 870,936,203
RAC: 1,552,924
Level
Glu
Scientific publications
wat
Message 55001 - Posted: 27 May 2020 | 23:02:12 UTC

I've been resorting to just resetting the project. Lost tasks be damned.

Richard Haselgrove
Send message
Joined: 11 Jul 09
Posts: 1003
Credit: 2,537,540,285
RAC: 3,303,364
Level
Phe
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 55004 - Posted: 28 May 2020 | 7:23:07 UTC

@Aurum: https solves nothing, but some posters at this and other projects feel that all internet traffic should be encrypted (it hardly seems necessary for BOINC data files). I have been suggesting editing the download urls to use http so that the work can be processed: Toni has suggested an alternative edit which might be more acceptable to some.

@mmonnin: basically, you're saying "other volunteers be damned" - if you don't process the task, somebody else has to cope with it.

But Toni has said that he's (hopefully) cured the problem at source, so let's hope we can stop talking about it.

Toni
Volunteer moderator
Project administrator
Project developer
Project scientist
Send message
Joined: 9 Dec 08
Posts: 947
Credit: 4,353,973
RAC: 48
Level
Ala
Scientific publications
watwatwatwat
Message 55005 - Posted: 28 May 2020 | 9:09:46 UTC - in response to Message 55004.

Feel free to abort any problematic WU. They should be re-created correct.

mmonnin
Send message
Joined: 2 Jul 16
Posts: 284
Credit: 870,936,203
RAC: 1,552,924
Level
Glu
Scientific publications
wat
Message 55006 - Posted: 28 May 2020 | 22:34:59 UTC - in response to Message 55004.

@Aurum: https solves nothing, but some posters at this and other projects feel that all internet traffic should be encrypted (it hardly seems necessary for BOINC data files). I have been suggesting editing the download urls to use http so that the work can be processed: Toni has suggested an alternative edit which might be more acceptable to some.

@mmonnin: basically, you're saying "other volunteers be damned" - if you don't process the task, somebody else has to cope with it.

But Toni has said that he's (hopefully) cured the problem at source, so let's hope we can stop talking about it.


If I abort it they will go to another person. If I reset the project the stay 'In progress' until they timeout. And then go to another person. There is nothing I can do to not make it go to someone else.

I cannot process a task the project will not upload to me. Aborting the task still says I have a task stuck in download and I cannot download another.

Keith Myers
Send message
Joined: 13 Dec 17
Posts: 508
Credit: 532,018,683
RAC: 1,714,122
Level
Lys
Scientific publications
wat
Message 55007 - Posted: 29 May 2020 | 1:41:11 UTC

Aborting the task still says I have a task stuck in download and I cannot download another.

Exit the client and Manager, pause a minute for pending operations to clear, once you have aborted a task to allow the client_state file to be updated or you will continue to get the "some task is stalled" message.

Aurum
Avatar
Send message
Joined: 12 Jul 17
Posts: 252
Credit: 9,791,563,847
RAC: 2,648,925
Level
Tyr
Scientific publications
wat
Message 55008 - Posted: 29 May 2020 | 11:42:46 UTC

Last couple of days I haven't seen a WU that won't move if kicked by a Retry All signal from BoincTasks. However, most WUs sit in my Transfer list both UL Pending or DL Pending until I manually Retry All. So every day I get up the majority of my computers are idle waiting for GG WUs to move. We went months with WUs flowing freely up & down and now almost every one gets stuck in the turnstall. Something changed that turned this into a babysitting required project.

Keith Myers
Send message
Joined: 13 Dec 17
Posts: 508
Credit: 532,018,683
RAC: 1,714,122
Level
Lys
Scientific publications
wat
Message 55009 - Posted: 29 May 2020 | 15:01:18 UTC

If you have more than a single computer attached to the project you will get periods of reduced connectivity because only one computer can be talking to the project at any one time. Until the connection to the project from one computer goes quiescent, then the other computers can't connect.

Even viewing the website in a browser is considered an active connection from your IP address and will prevent a computer from performing an upload or download.

There is a dedicated thread on this issues that hasn't received a remedial response from Toni yet.

https://www.gpugrid.net/forum_thread.php?id=5127

Post to thread

Message boards : Server and website : Problem resolved: new WUs stuck in "downloading" status