stuck download disaster

Message boards : Number crunching : stuck download disaster
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile JStateson
Volunteer tester
Avatar

Send message
Joined: 27 May 99
Posts: 120
Credit: 22,844,504
RAC: 50,763
United States
Message 1989198 - Posted: 7 Apr 2019, 20:53:55 UTC
Last modified: 7 Apr 2019, 21:41:22 UTC

I was messing around with an old core 2 quad motherboard that had four x1 slots and one x16 slot. Fresh install of windows 10 and with five rx-560s.

All seemed fine***, I had 4 seti v8 ati tasks running and messages said it was too soon to get additional tasks for the other GPU. I then went to my account and allowed CPU tasks in addition to GPU. I set ngpu to 1.0 and ncpu to 0.4 so as to free up exactly 2 cpus for seti cpu work units.

I got over 100 cpu tasks downloaded all of which were stuck in transfer. CPU and GPU usage showed nothing happening. After a few minutes of nothing happening, I issued a --quit to boinccmd. That did not work so I rebooted.

Reboot did not solve any of the stuck transfers, I aborted all the transfers and reset the project. Right now I have 5 collatz running at %10 priority and hopefully seti at %100 will change its mind and download some more. Last I looked I saw "last request too recent"

***was not fine at first but that is subject of another post.
ID: 1989198 · Report as offensive     Reply Quote
Profile Tom M
Volunteer tester

Send message
Joined: 28 Nov 02
Posts: 3042
Credit: 132,392,632
RAC: 759,261
United States
Message 1989268 - Posted: 8 Apr 2019, 12:14:49 UTC - in response to Message 1989198.  
Last modified: 8 Apr 2019, 12:15:54 UTC

I was messing around with an old core 2 quad motherboard that had four x1 slots and one x16 slot. Fresh install of windows 10 and with five rx-560s.

All seemed fine***, I had 4 seti v8 ati tasks running and messages said it was too soon to get additional tasks for the other GPU. I then went to my account and allowed CPU tasks in addition to GPU. I set ngpu to 1.0 and ncpu to 0.4 so as to free up exactly 2 cpus for seti cpu work units.

I got over 100 cpu tasks downloaded all of which were stuck in transfer. CPU and GPU usage showed nothing happening. After a few minutes of nothing happening, I issued a --quit to boinccmd. That did not work so I rebooted.

Reboot did not solve any of the stuck transfers, I aborted all the transfers and reset the project. Right now I have 5 collatz running at %10 priority and hopefully seti at %100 will change its mind and download some more. Last I looked I saw "last request too recent"

***was not fine at first but that is subject of another post.


We all have had trouble with stuck downloads. It is not the fault of the way you are setup.

There are a couple of things you can try. Select the last of the stuck downloads and hit the retry button.

If you are running multiple projects on a box, you can also change the maximum file transfers to say, 16 and the maximum project transfers to say 8. That way, when Seti gets stuck (again) the other project will not get hung.

Another thing to do is put an upper limit on the "reporting tasks", say 100 or 250. This will help in that when you report a lot of tasks after a Seti outage (and/or on tuesday) it doesn't jam up your reporting. Until you have reported successfully, Seti will not download more tasks.

Sometimes you have to do a manual update to get it to "report".

When you run a lot of gpu's the behavior of the uploads/downloads reliability is more cranky.

Tom
I will stop procrastinating tomorrow.
\\// Live Long & Prosper (starting tomorrow ;)
ID: 1989268 · Report as offensive     Reply Quote
Profile JStateson
Volunteer tester
Avatar

Send message
Joined: 27 May 99
Posts: 120
Credit: 22,844,504
RAC: 50,763
United States
Message 1989286 - Posted: 8 Apr 2019, 14:09:25 UTC - in response to Message 1989268.  



We all have had trouble with stuck downloads. It is not the fault of the way you are setup.

There are a couple of things you can try. Select the last of the stuck downloads and hit the retry button.

If you are running multiple projects on a box, you can also change the maximum file transfers to say, 16 and the maximum project transfers to say 8. That way, when Seti gets stuck (again) the other project will not get hung.

Another thing to do is put an upper limit on the "reporting tasks", say 100 or 250. This will help in that when you report a lot of tasks after a Seti outage (and/or on tuesday) it doesn't jam up your reporting. Until you have reported successfully, Seti will not download more tasks.

Sometimes you have to do a manual update to get it to "report".

When you run a lot of gpu's the behavior of the uploads/downloads reliability is more cranky.

Tom


was not aware of these options. Normally I check to see if there is any bytes/sec at all. If not then I retry the one at the top unless it has a humongous size. Setting max number of files is better. This system had only 4 cores of which all were used %100 which contributed to the problem.
ID: 1989286 · Report as offensive     Reply Quote

Message boards : Number crunching : stuck download disaster


 
©2019 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.