no new tasks, no results online and task saves / checkpoints not working

Message boards : Number crunching : no new tasks, no results online and task saves / checkpoints not working

To post messages, you must log in.

AuthorMessage
gazzawazza

Send message
Joined: 4 May 07
Posts: 28
Credit: 297,648
RAC: 0
Message 74089 - Posted: 23 Oct 2012, 22:34:29 UTC

********* Re-posted from Questions and Answers: Windows Section ********

hi all

I don't know whether these are connected but I'm experiencing some problems with Rosetta.

I've taken a big break from crunching and on return my BOINC client (7.0.28) started high priority crunching on a couple of tasks which go back to May 2012.

FIRST PROBLEM:
These tasks don't appear to be saving I.e. the tasks don't seem to be generating checkpoints. I did some digging on the subject and only found old posts indicating that save-points should occur on every percentage completed.

As this doesn't seem to be happening, I've aborted the tasks.

SECOND PROBLEM:
However, before I aborted the tasks, I tried to find any insight into the problem and was checking out my results. Unfortunately, when I tried to do this, I only got the following error:

"Sorry, the data requested does not exist."

As I have 97005 credits, I would have thought some results should have appeared here.

Finally, I did read on the Rosetta homepage that you've been doing a network upgrade, so thought this might conceivably explain the results error but can't see how this would affect the generation of checkpoints (which I'm pretty sure is simply a localised action).

THIRD PROBLEM:
I've reset the project and tried to get new tasks but nothing's been downloaded to my BOINC client.

Any suggestions?



Thanks,

Gary

PS sorry if I'm being a noob :)
ID: 74089 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Polian
Avatar

Send message
Joined: 21 Sep 05
Posts: 152
Credit: 10,141,266
RAC: 0
Message 74090 - Posted: 24 Oct 2012, 4:30:28 UTC - in response to Message 74089.  

********* Re-posted from Questions and Answers: Windows Section ********

hi all

I don't know whether these are connected but I'm experiencing some problems with Rosetta.

I've taken a big break from crunching and on return my BOINC client (7.0.28) started high priority crunching on a couple of tasks which go back to May 2012.

FIRST PROBLEM:
These tasks don't appear to be saving I.e. the tasks don't seem to be generating checkpoints. I did some digging on the subject and only found old posts indicating that save-points should occur on every percentage completed.

As this doesn't seem to be happening, I've aborted the tasks.

SECOND PROBLEM:
However, before I aborted the tasks, I tried to find any insight into the problem and was checking out my results. Unfortunately, when I tried to do this, I only got the following error:

"Sorry, the data requested does not exist."

As I have 97005 credits, I would have thought some results should have appeared here.

Finally, I did read on the Rosetta homepage that you've been doing a network upgrade, so thought this might conceivably explain the results error but can't see how this would affect the generation of checkpoints (which I'm pretty sure is simply a localised action).

THIRD PROBLEM:
I've reset the project and tried to get new tasks but nothing's been downloaded to my BOINC client.

Any suggestions?



Thanks,

Gary

PS sorry if I'm being a noob :)


Hi Gary
Yes, not only can you abort the May tasks, you should abort them. They have already been recorded as no reply, sent to a secondary cruncher, returned, processed, and removed from the database.

You don't see anything when you try and look at the task history because all the results from when you were last active were already removed from the BOINC database. (working as intended)

Odd that after a project reset it wouldn't get new tasks. Have you exited and restarted BOINC?

and lol@ noob comment, no worries.
ID: 74090 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Mod.Sense
Volunteer moderator

Send message
Joined: 22 Aug 06
Posts: 4018
Credit: 0
RAC: 0
Message 74095 - Posted: 25 Oct 2012, 1:52:59 UTC

Please post the messages you get when you do an update to the project in the BOINC Manager. Have you modified any of the configuration settings? Are you running any other BOINC projects?
Rosetta Moderator: Mod.Sense
ID: 74095 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Link
Avatar

Send message
Joined: 4 May 07
Posts: 352
Credit: 382,349
RAC: 0
Message 74098 - Posted: 25 Oct 2012, 19:08:45 UTC - in response to Message 74091.  

Checkpoints: They are only guaranteed to occur when a model within a task is generated. As far as I know the percentage of a task completed doesn't guarantee a checkpoint.

And that means, that it can take even several hours, before the task checkpoints. Set your BOINC preferences to run always, leave applications in memory when suspended and do not shut down your computer too often, hibernate it instead. That way you should not loose too much work.
.
ID: 74098 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
gazzawazza

Send message
Joined: 4 May 07
Posts: 28
Credit: 297,648
RAC: 0
Message 74099 - Posted: 25 Oct 2012, 19:48:39 UTC - in response to Message 74090.  

Hi all

Thanks for all the replies :)


@Polian

You don't see anything when you try and look at the task history because all the results from when you were last active were already removed from the BOINC database. (working as intended)


Ah... so you can only review a limited history of tasks? What's the timeframe? 30 days or something?

Odd that after a project reset it wouldn't get new tasks. Have you exited and restarted BOINC?


Yep, tried that 2 or 3 times (over the last couple of days) and tried updating the project too




@Transient

Checkpoints: They are only guaranteed to occur when a model within a task is generated. As far as I know the percentage of a task completed doesn't guarantee a checkpoint.


This is the link that indicated about percentage checkpoints. Obviously it's rather old:

https://boinc.bakerlab.org/rosetta/forum_thread.php?id=669&nowrap=true#10374




@Mod.Sense

Please post the messages you get when you do an update to the project in the BOINC Manager. Have you modified any of the configuration settings? Are you running any other BOINC projects?


Here's a recent extract from my BOINC client log:

"25/10/2012 20:26:48 | rosetta@home | update requested by user
25/10/2012 20:26:53 | rosetta@home | Sending scheduler request: Requested by user.
25/10/2012 20:26:53 | rosetta@home | Not reporting or requesting tasks
25/10/2012 20:26:55 | rosetta@home | Scheduler request completed"

I'm not aware of having made any alterations to my configuration files. Certainly nothing specific to Rosetta (at least that I'm aware of). I know I logged into my account, when visiting this message board, but that was obviously after I'd been experiencing these issues.

And yes, I'm running a number of projects (and have done for some time, generally, without incident):

Climateprediction.net
Malariacontrol.net
rosetta@home
SETI@home
World Community Grid

I'm currently crunching tasks from all but Rosetta.




Regards,

Gary
ID: 74099 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Mod.Sense
Volunteer moderator

Send message
Joined: 22 Aug 06
Posts: 4018
Credit: 0
RAC: 0
Message 74101 - Posted: 26 Oct 2012, 0:17:45 UTC
Last modified: 26 Oct 2012, 0:19:27 UTC

25/10/2012 20:26:53 | rosetta@home | Not reporting or requesting tasks


This is saying that the BOINC Manager does not feel it will have time to process any new tasks from Rosetta at this time, and so it is not requesting any new work. BOINC attempts to estimate the average amount of time per day your machine is running BOINC, and other factors, and requests new work when it is fairly confident the work can be completed on time. And so at this time, it does not have that confidence. This can sometimes be due to tasks which poorly estimate what their completion times will be (i.e. the time remaining). Or due to recent changes in your habits with how you use the machine (if you begin running 24hrs a day rather than 12 for example, or visa-versa).

Bottom line, the underlaying cause of you not receiving any work is that you are not requesting any. Once the other tasks you already have get further along to completion, it will request work from Rosetta.
Rosetta Moderator: Mod.Sense
ID: 74101 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
gazzawazza

Send message
Joined: 4 May 07
Posts: 28
Credit: 297,648
RAC: 0
Message 74109 - Posted: 26 Oct 2012, 13:43:21 UTC - in response to Message 74101.  

25/10/2012 20:26:53 | rosetta@home | Not reporting or requesting tasks


This is saying that the BOINC Manager does not feel it will have time to process any new tasks from Rosetta at this time, and so it is not requesting any new work. BOINC attempts to estimate the average amount of time per day your machine is running BOINC, and other factors, and requests new work when it is fairly confident the work can be completed on time. And so at this time, it does not have that confidence. This can sometimes be due to tasks which poorly estimate what their completion times will be (i.e. the time remaining). Or due to recent changes in your habits with how you use the machine (if you begin running 24hrs a day rather than 12 for example, or visa-versa).

Bottom line, the underlaying cause of you not receiving any work is that you are not requesting any. Once the other tasks you already have get further along to completion, it will request work from Rosetta.


Hi Mod.sense

Thanks for the explanation. I swear I read that the way new tasks are requested has been recoded at some point in the 7.x clients and that BOINC effectively has to establish baselines (a high and low threshold of usage) before requesting new work?

If so, what I don't understand is how / why I've received, completed & returned tasks for all the other projects I'm associated with in the last few days, aside from Rosetta, despite basically not having been crunching at all since around April? If I understand the new scheduler process (even a little), I'd have thought that, having defaulted on tasks that I was issued with 6 months ago, would have created a dip in my projected ability to complete work I.e. I'd have thought my 'capacity' to complete tasks would be at issue with all the projects, not just Rosetta.

Having said all this, I know that I'm dumbing down the scheduler process significantly and there are multiple factors being considered before new tasks are issued.

I understand there's a more verbose log available (debug). Is it worth running this to try and establish why my client reckons it can't handle new Rosetta tasks at the moment?

Also, let's say, for arguments sake, that my thresholds for Rosetta are not accurate - can I reset them?


Regards,

Gary
ID: 74109 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
gazzawazza

Send message
Joined: 4 May 07
Posts: 28
Credit: 297,648
RAC: 0
Message 74143 - Posted: 31 Oct 2012, 16:40:30 UTC - in response to Message 74118.  

You could try a detach/attach, or a reset of Rosetta from the projects tab.

Edit: I see you got a new result in. Hopefully everything will go smoothly now.


Hi Transient

Thanks for your kindness in replying and keeping an eye on things. Yeah, am now getting Rosetta tasks again. Seems like BONIC was just thinking about whether it could crunch Rosetta as well as all the other projects / tasks I'm doing.

Shame though that it took a week for my client to decide this - would like to think I'd be forgiven for wondering why I wasn't getting new tasks.

I think I read that a new build of BOINC is going to include reasons for not getting new tasks, which I think may mop up some queries (such as my one). As per normal, I can't find the the page where I read this lol.


Cheers,

Gary
ID: 74143 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote

Message boards : Number crunching : no new tasks, no results online and task saves / checkpoints not working



©2024 University of Washington
https://www.bakerlab.org