no progress

LFenske
LFenske
Joined: 2 Apr 09
Posts: 9
Credit: 279084
RAC: 0
Topic 194602

Some of my Einstein@Home tasks stop making progress. After a while, neither the "CPU time" nor "Progress" fields change. I have to end up aborting those tasks. I'm running Ubuntu 8.04 64-bit with BOINC 5.10.45 (the latest available through the repository). I have four other BOINC projects that work, so I'm ready to give up on Einstein@Home.

Gary Roberts
Gary Roberts
Moderator
Joined: 9 Feb 05
Posts: 5848
Credit: 110008249276
RAC: 24424407

no progress

Quote:
Some of my Einstein@Home tasks stop making progress. After a while, neither the "CPU time" nor "Progress" fields change. I have to end up aborting those tasks. I'm running Ubuntu 8.04 64-bit with BOINC 5.10.45 (the latest available through the repository). I have four other BOINC projects that work, so I'm ready to give up on Einstein@Home.


Your list of computers shows just one machine attached to E@H. The tasks list shows some successfully completed tasks and just one that you aborted.

Are you running the four other projects on this machine as well as E@H? If you were, it would be quite normal to see E@H tasks paused and making no further progress while the tasks of other projects were given their turn. Is the E@H task actually listed as 'running' in BOINC Manager when it appears to be making no progress? As your machine is a dual core, you should see two running tasks and you may also see other tasks 'waiting to run'. How often this changes is controlled by what task switch interval you have set. The default value is one hour. If the problem is something else, please give more details.

Cheers,
Gary.

Gundolf Jahn
Gundolf Jahn
Joined: 1 Mar 05
Posts: 1079
Credit: 341280
RAC: 0

RE: Your list of computers

Message 95357 in response to message 95356

Quote:
Your list of computers shows just one machine attached to E@H.


And I'd like to add that your Einstein account doesn't show any other projects listed.

Quote:
The tasks list shows some successfully completed tasks and just one that you aborted.


However, the two S5R5 tasks (when I looked yesterday, it's been three) show a behaviour like those infamous "too many exits" tasks, just not enough to be aborted. So, are you throttling your CPU, i.e. do you allow less than 100% of CPU time being used?

Gruß,
Gundolf

Computer sind nicht alles im Leben. (Kleiner Scherz)

LFenske
LFenske
Joined: 2 Apr 09
Posts: 9
Credit: 279084
RAC: 0

RE: RE: Some of my

Message 95358 in response to message 95356

Quote:
Quote:
Some of my Einstein@Home tasks stop making progress. After a while, neither the "CPU time" nor "Progress" fields change. I have to end up aborting those tasks. I'm running Ubuntu 8.04 64-bit with BOINC 5.10.45 (the latest available through the repository). I have four other BOINC projects that work, so I'm ready to give up on Einstein@Home.

Your list of computers shows just one machine attached to E@H. The tasks list shows some successfully completed tasks and just one that you aborted.

Are you running the four other projects on this machine as well as E@H? If you were, it would be quite normal to see E@H tasks paused and making no further progress while the tasks of other projects were given their turn. Is the E@H task actually listed as 'running' in BOINC Manager when it appears to be making no progress? As your machine is a dual core, you should see two running tasks and you may also see other tasks 'waiting to run'. How often this changes is controlled by what task switch interval you have set. The default value is one hour. If the problem is something else, please give more details.

I apologize for not being clearer from the beginning.

Yes, all five projects are on this one machine.

Yes, the Einstein@Home task is actually listed as 'running' when it appears to be making no progress, along with one other task, since this is dual core. Einstein@Home is allocated a 5% resource share. In the past, I have let it go for a week, and seen no progress. Also, when a task is in this state, xload shows a load average of only around one (the other running task). When it does work, the load average is around two. It never gets stuck at the very beginning or very end of the task, but when it happens it is always with some fraction of the task finished. I have also had this happen with some World Community Grid projects, but never with SETI@home, ClimatePrediction, or Milkyway@home. I often have a dozen tasks queued of various kinds, so I have seen it rotate through tasks from the five projects.

Thank you for your reply. If you want any more information, I'm happy to comply.

LFenske
LFenske
Joined: 2 Apr 09
Posts: 9
Credit: 279084
RAC: 0

RE: RE: Your list of

Message 95359 in response to message 95357

Quote:
Quote:
Your list of computers shows just one machine attached to E@H.

And I'd like to add that your Einstein account doesn't show any other projects listed.

Quote:
The tasks list shows some successfully completed tasks and just one that you aborted.

However, the two S5R5 tasks (when I looked yesterday, it's been three) show a behaviour like those infamous "too many exits" tasks, just not enough to be aborted. So, are you throttling your CPU, i.e. do you allow less than 100% of CPU time being used?

Gruß,
Gundolf

The other projects I have running are not Einstein@home projects (I didn't know there were more).

I'm not familiar with the infamous "too many exits" problems.

I am not throttling my CPU. I allow 100% of the processors and 100% of CPU time.

I'm happy to answer more questions to get this resolved.

Gundolf Jahn
Gundolf Jahn
Joined: 1 Mar 05
Posts: 1079
Credit: 341280
RAC: 0

RE: The other projects I

Message 95360 in response to message 95359

Quote:
The other projects I have running are not Einstein@home projects (I didn't know there were more).


There aren't. I was speaking of projects like SETI@home, ClimatePrediction or Milkyway@home. It's strange that they don't show up at your account. Click on my name (or Gary's) in the Author column to see of what I'm speaking.

Quote:

I'm not familiar with the infamous "too many exits" problems.

I am not throttling my CPU. I allow 100% of the processors and 100% of CPU time.


I was only guessing there, since your tasks are frequently restarted within a few minutes. If that would happen more than 100 times in succession, the task would be aborted with that error message. Oftentimes CPU throttling is causing that behaviour, that's why I asked.

Did you try to suspend/resume the task or BOINC as a whole to "unstuck" the task? Does a reboot help?

By the way, I'm running version 5.10.45 too (on windows XP), without any problems. (And 5.8.16 on NT4, but not with Einstein :-)

Gruß,
Gundolf

Computer sind nicht alles im Leben. (Kleiner Scherz)

Gary Roberts
Gary Roberts
Moderator
Joined: 9 Feb 05
Posts: 5848
Credit: 110008249276
RAC: 24424407

RE: Yes, the Einstein@Home

Message 95361 in response to message 95358

Quote:
Yes, the Einstein@Home task is actually listed as 'running' when it appears to be making no progress....


Is your machine overclocked? I run a large number of machines, most of which are overclocked and on older ones, where the cooling may be less efficient than it used to be, I have seen a similar behaviour, particularly if the ambient is rather warm. The symptoms are pretty much as you describe. The machine doesn't lock up or crash but an E@H task (supposedly running) does cease to make any progress. I could always kick start the task (at least temporarily) by simply stopping and restarting BOINC. Have you specifically tried that? You really don't need to abort them as changing to a different task is unlikely to change the behaviour. The problem is most likely to result from some hardware issue on your machine.

E@H tasks do seem to be harder on the processor than those of many other projects. Whenever I found this type of problem, I could always remove it by backing off on the overclock and by lowering the ambient or cleaning the heat sink and checking the fan to make sure CPU cooling was as good as possible.

Quote:
I have also had this happen with some World Community Grid projects, but never with SETI@home....


I checked over at Seti for the name 'LFenske' and drew a blank. I guess you must be using different account names at different projects and this would be why your other projects are not listed at the bottom of your user account information here, as Gundolf was talking about. There is no particular problem with using different account details at different projects - other than not being able to see all your stats in the one place.

Cheers,
Gary.

Gundolf Jahn
Gundolf Jahn
Joined: 1 Mar 05
Posts: 1079
Credit: 341280
RAC: 0

RE: I checked over at Seti

Message 95362 in response to message 95361

Quote:
I checked over at Seti for the name 'LFenske' and drew a blank. I guess you must be using different account names at different projects and this would be why your other projects are not listed at the bottom of your user account information here, as Gundolf was talking about. There is no particular problem with using different account details at different projects - other than not being able to see all your stats in the one place.


That shouldn't depend on the user name but on the email address, since the Cross Project ID (the common denominator) is derived from that (except for WCG).

Gruß,
Gundolf

Computer sind nicht alles im Leben. (Kleiner Scherz)

LFenske
LFenske
Joined: 2 Apr 09
Posts: 9
Credit: 279084
RAC: 0

RE: RE: I checked over at

Message 95363 in response to message 95362

Quote:
Quote:
I checked over at Seti for the name 'LFenske' and drew a blank. I guess you must be using different account names at different projects and this would be why your other projects are not listed at the bottom of your user account information here, as Gundolf was talking about. There is no particular problem with using different account details at different projects - other than not being able to see all your stats in the one place.

That shouldn't depend on the user name but on the email address, since the Cross Project ID (the common denominator) is derived from that (except for WCG).

Gruß,
Gundolf

Ah, yes, I use a different e-mail address for nearly everything I do; my five BOINC projects are all under different e-mail addresses, plus another for BOINC itself. That may have been a mistake. I didn't know they could be correlated by e-mail address. I wonder if I can consolidate them at this point.

Jord
Joined: 26 Jan 05
Posts: 2952
Credit: 5779100
RAC: 0

RE: I didn't know they

Message 95364 in response to message 95363

Quote:
I didn't know they could be correlated by e-mail address. I wonder if I can consolidate them at this point.


Accounts are registered to one unique thing: the email address. You can change the email address on the projects to get them all together, no need to go make new accounts.

Just go to any of the projects "Your account" pages and change the email address. You will need the password per project to be able to change that.

For Einstein, that's here.

Gundolf Jahn
Gundolf Jahn
Joined: 1 Mar 05
Posts: 1079
Credit: 341280
RAC: 0

RE: I wonder if I can

Message 95365 in response to message 95363

Quote:
I wonder if I can consolidate them at this point.


That should be no problem. Just go to your account at each project (but one :-) and change the email address. That'll create new CPIDs at first, but they will align with time.

Gruß,
Gundolf
[edit]Okay, late again :-)[/edit]

Computer sind nicht alles im Leben. (Kleiner Scherz)

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.