Welcome to MilkyWay@home

Tasks disappear from machine

Questions and Answers : Unix/Linux : Tasks disappear from machine
Message board moderation

To post messages, you must log in.

AuthorMessage
Ralph Little

Send message
Joined: 30 Jul 16
Posts: 6
Credit: 58,669,504
RAC: 0
Message 69978 - Posted: 7 Jul 2020, 5:33:04 UTC

I'm been running GPU work through a couple of machines that I have.

One of them (id=764322) is pumping through GPU work and I see them in the task view for the machine.
One of the others (id=822434), is also pumping through pure GPU work and sending it up. However, for that machine, the jobs quickly disappear from my job list leaving only a very small number remaining.

Both machines are running Linux Mint 19.3.

The first is a laptop with a different NVidia GPU.
The GPU on the second machine is a GTX 1660 (stock everything, no overclocking).

Anybody have any idea what is going on here? Is the work getting rejected for some reason?
ID: 69978 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile mikey
Avatar

Send message
Joined: 8 May 09
Posts: 3315
Credit: 519,950,508
RAC: 21,896
Message 69979 - Posted: 7 Jul 2020, 10:59:44 UTC - in response to Message 69978.  

I'm been running GPU work through a couple of machines that I have.

One of them (id=764322) is pumping through GPU work and I see them in the task view for the machine.
One of the others (id=822434), is also pumping through pure GPU work and sending it up. However, for that machine, the jobs quickly disappear from my job list leaving only a very small number remaining.

Both machines are running Linux Mint 19.3.

The first is a laptop with a different NVidia GPU.
The GPU on the second machine is a GTX 1660 (stock everything, no overclocking).

Anybody have any idea what is going on here? Is the work getting rejected for some reason?


Both machines are crunching just fine, the gpu jobs are much shorter than cpu jobs, mine are around a minute because I use older AMD gpu's, so will 'disappear' quicker from the machine. But if you look at your computers both of them are doing very well. You can see your computers here and then you can click on Tasks and see how each is doing:
[url]https://milkyway.cs.rpi.edu/milkyway/hosts_user.php?userid=1116944[/url[
ID: 69979 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Ralph Little

Send message
Joined: 30 Jul 16
Posts: 6
Credit: 58,669,504
RAC: 0
Message 69980 - Posted: 7 Jul 2020, 14:13:05 UTC - in response to Message 69979.  
Last modified: 7 Jul 2020, 14:18:48 UTC

I see the credit figures and they do look reasonable.

However, if you click through to the actual tasks by machine, you will see that "rlittlelaptop" (764322) has a large number of tasks listed, whereas "littleserver1" (822434) isn't even listing any tasks for the 6th or 7th of June.
If I watch the tasks completed, I see them transition to the task page as "completed" but they soon disappear from the machine's list, in a way that the tasks don't for rlittlelaptop.

I should clarify that when I mean disappear from the machine's list, I mean on the website. The GPU jobs take about 4mins each on the machines themselves.

I just wanted to make sure they were not being purged due to computation error.

I was worried about the apparent discrepancy at the server side.
ID: 69980 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile mikey
Avatar

Send message
Joined: 8 May 09
Posts: 3315
Credit: 519,950,508
RAC: 21,896
Message 69982 - Posted: 7 Jul 2020, 23:36:53 UTC - in response to Message 69980.  

I see the credit figures and they do look reasonable.

However, if you click through to the actual tasks by machine, you will see that "rlittlelaptop" (764322) has a large number of tasks listed, whereas "littleserver1" (822434) isn't even listing any tasks for the 6th or 7th of June.
If I watch the tasks completed, I see them transition to the task page as "completed" but they soon disappear from the machine's list, in a way that the tasks don't for rlittlelaptop.

I should clarify that when I mean disappear from the machine's list, I mean on the website. The GPU jobs take about 4mins each on the machines themselves.

I just wanted to make sure they were not being purged due to computation error.

I was worried about the apparent discrepancy at the server side.


On your list of tasks it will list the ones with errors or computation errors and neither machine shows any problems with any of the tasks.
BTW only you can see the actual machine names all we see is them sorted by the last time they connected to the project and a number 1, 2 etc. If you click on my name you can see my computers but not the actual names of the pc's just numbers.
ID: 69982 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Ralph Little

Send message
Joined: 30 Jul 16
Posts: 6
Credit: 58,669,504
RAC: 0
Message 69983 - Posted: 8 Jul 2020, 1:23:19 UTC
Last modified: 8 Jul 2020, 1:24:31 UTC

My mistake.

The littleserver1 was running a newer boinc than the rlittlelaptop machine and for some reason it had a problem with coming out of suspend when the machine was inactive.
When I was testing it, I had the machine set to run GPU tasks all the time whereas my preferences were to have it suspended on activity.

I downgraded the boinc and it seems to be working as normal now.
ID: 69983 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile mikey
Avatar

Send message
Joined: 8 May 09
Posts: 3315
Credit: 519,950,508
RAC: 21,896
Message 69984 - Posted: 8 Jul 2020, 10:17:14 UTC - in response to Message 69983.  
Last modified: 8 Jul 2020, 10:20:56 UTC

My mistake.

The littleserver1 was running a newer boinc than the rlittlelaptop machine and for some reason it had a problem with coming out of suspend when the machine was inactive.
When I was testing it, I had the machine set to run GPU tasks all the time whereas my preferences were to have it suspended on activity.

I downgraded the boinc and it seems to be working as normal now.


Those settings can also be adjusted within the Boinc Manager itself, the Boinc Manager is down by the clock and if you click the up arrow you can see it, then right click to bring up some options then left click on 'open Boinc Manager' Boinc will open up for you. After that click on Options, computing preferences and the first tab, computing, in the 2nd section is all about when to suspend Boinc. Doing it this way is a computer by computer setting but you can also do it for every pc if you click on your name at the top right of this page and then in the 2nd section. preferences, choose the top line 'computing preferences' and make your changes there.

The choices of when to suspend when the computer is busy are the default settings in every version of Boinc, the newer versions tend to fix problems the older versions had like most recently an internet security certificate needed updating so they included it in with a few fairly minor changes. But if a version works for you you do not need to update just because there is a new version out there.
ID: 69984 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Ralph Little

Send message
Joined: 30 Jul 16
Posts: 6
Credit: 58,669,504
RAC: 0
Message 69985 - Posted: 8 Jul 2020, 14:22:04 UTC - in response to Message 69984.  
Last modified: 8 Jul 2020, 14:23:27 UTC

Yeah, I do remember updating Boinc on this machine using a PPA for some reason.
The reason now escapes me.
I think it was something to do with Rossetta, perhaps graphics (which I don't really care about anyway) but I forget. :)

Everything is good now. 99 mostly completed tasks on the server this morning against the previously errant machine.

Thanks for the replies!
ID: 69985 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile mikey
Avatar

Send message
Joined: 8 May 09
Posts: 3315
Credit: 519,950,508
RAC: 21,896
Message 69986 - Posted: 8 Jul 2020, 20:25:47 UTC - in response to Message 69985.  

Yeah, I do remember updating Boinc on this machine using a PPA for some reason.
The reason now escapes me.
I think it was something to do with Rossetta, perhaps graphics (which I don't really care about anyway) but I forget. :)

Everything is good now. 99 mostly completed tasks on the server this morning against the previously errant machine.

Thanks for the replies!


No problem just glad it's all working now
ID: 69986 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote

Questions and Answers : Unix/Linux : Tasks disappear from machine

©2024 Astroinformatics Group