log in

Advanced search

Message boards : News : Temporary Suspension of Work

Author Message
Profile Tom
Volunteer moderator
Avatar
Send message
Joined: 23 Jun 08
Posts: 490
Credit: 238,767
RAC: 0
Message 3204 - Posted: 25 Jun 2014, 13:46:04 UTC
Last modified: 26 Jun 2014, 19:30:48 UTC

Hi all,

We're temporarily suspending our BOINC feeder while we deploy new versions of all our applications. It shouldn't take more than a couple of hours. I'll keep you posted.

Cheers,
-Tom

Profile Aurel
Send message
Joined: 25 Feb 13
Posts: 36
Credit: 47,446
RAC: 0
Message 3205 - Posted: 25 Jun 2014, 15:17:58 UTC - in response to Message 3204.

Hi all,

We're temporarily suspending our BOINC feeder while we deploy new versions of all our applications. It shouldn't take more than a couple of hours. I'll keep you posted.

Cheers,


Hi,

ok, thats the reasen why I only get "feeder not running" in the event log. :) What will bring the updates?`
Is there something huge on the way???
Or is it because pypy 2.3.1 can handle with python?

While we support 32 bit python on Windows, work on the native Windows 64 bit python is still stalling, we would welcome a volunteer to handle that.

Profile Tom
Volunteer moderator
Avatar
Send message
Joined: 23 Jun 08
Posts: 490
Credit: 238,767
RAC: 0
Message 3207 - Posted: 25 Jun 2014, 21:10:21 UTC
Last modified: 26 Jun 2014, 1:15:22 UTC

Aurel, I'll have more information available tomorrow. For right now, I think I have things working to continue crunching throughout the night, but please report any problems you're experiencing in the meantime.

So far I've fixed one bug related to the download link (for applications and input files) and another related to a path issue for the CCL app (the actual error results in a "File Not Found").

Cheers,

Profile Tom
Volunteer moderator
Avatar
Send message
Joined: 23 Jun 08
Posts: 490
Credit: 238,767
RAC: 0
Message 3213 - Posted: 26 Jun 2014, 19:25:40 UTC

So, the latest release from the last couple of days had several updates:


  • New versions of the 7za unzip tool (including true 64-bit binaries for Linux users)
  • 64-bit versions of each Linux Application
  • New versions of the R application that fixes a permissions issue problematic on some hosts
  • Server updates that will evenly distribute work between jobs (this prevents platform-dependent tasks from filling up and blocking the work queues when cross-compatible tasks are available)


This update address several long standing issues reported in the following threads:

Grandpa
Send message
Joined: 6 Jun 14
Posts: 2
Credit: 1,783,613
RAC: 0
Message 3223 - Posted: 28 Jun 2014, 22:39:29 UTC - in response to Message 3213.

So, the latest release from the last couple of days had several updates:

  • New versions of the 7za unzip tool (including true 64-bit binaries for Linux users)
  • 64-bit versions of each Linux Application
  • New versions of the R application that fixes a permissions issue problematic on some hosts
  • Server updates that will evenly distribute work between jobs (this prevents platform-dependent tasks from filling up and blocking the work queues when cross-compatible tasks are available)


This update address several long standing issues reported in the following threads:



I do not know which of the above issues I was experiencing before but previously out of 7 Linux 64bit 4 processor machines I have, I had one Linux 64bit box that would not run MindMolding. I am happy to report that all of the boxen are now able to run it.
Thanks for the fixes.

Grandpa
Send message
Joined: 6 Jun 14
Posts: 2
Credit: 1,783,613
RAC: 0
Message 3224 - Posted: 28 Jun 2014, 23:23:50 UTC - in response to Message 3223.
Last modified: 28 Jun 2014, 23:24:25 UTC

I have seen some rather strange behaviour in a few of the WU's. the 11 listed below I had on 1 of the computers and I noticed that after around 50 min. they still showed 0% progress so I aborted them. I notced when checking the stats / my computers that they were all assigned to other machines I have running. I check them on the other computers and they were having the same behaviour 0% progress after very long run times. At the time of writing this some of them have completed some are still running the ones that have completed have very long run times 4x longer than normal and when they do complete they say 0% progress. I am however receiving appropriate credit for them for the amount of run time. But it seems there may be a bug of some kind since they have very long run times and show no progress in boinc manager. If they had not been assigned to several different computers of my own both Opteron and Xeon rigs I would not have noticed this, but since I did I figured I should report it.

17589307 16031688 28 Jun 2014, 18:42:24 UTC 28 Jun 2014, 19:31:46 UTC Aborted by user 2,860.01 2,838.53 --- ACT-R cognitive modeling environment using Clozure Common Lisp (Linux Only) v2.32 (sse2) 17589306 16031687 28 Jun 2014, 18:42:24 UTC 28 Jun 2014, 19:31:46 UTC Aborted by user 2,860.01 2,840.71 --- ACT-R cognitive modeling environment using Clozure Common Lisp (Linux Only) v2.32 (sse2) 17589305 16031686 28 Jun 2014, 18:40:19 UTC 28 Jun 2014, 19:31:46 UTC Aborted by user 2,984.65 2,969.71 --- ACT-R cognitive modeling environment using Clozure Common Lisp (Linux Only) v2.32 (sse2) 17589304 16031685 28 Jun 2014, 18:40:19 UTC 28 Jun 2014, 19:31:46 UTC Aborted by user 2,985.73 2,972.01 --- ACT-R cognitive modeling environment using Clozure Common Lisp (Linux Only) v2.32 (sse2) 17589303 16031684 28 Jun 2014, 18:39:07 UTC 28 Jun 2014, 19:31:46 UTC Aborted by user 3,056.70 3,041.65 --- ACT-R cognitive modeling environment using Clozure Common Lisp (Linux Only) v2.32 (sse2) 17589302 16031683 28 Jun 2014, 18:42:12 UTC 28 Jun 2014, 19:31:46 UTC Aborted by user 2,873.06 2,851.22 --- ACT-R cognitive modeling environment using Clozure Common Lisp (Linux Only) v2.32 (sse2) 17589301 16031682 28 Jun 2014, 18:39:19 UTC 28 Jun 2014, 19:31:46 UTC Aborted by user 3,033.70 3,018.81 --- ACT-R cognitive modeling environment using Clozure Common Lisp (Linux Only) v2.32 (sse2) 17589300 16031681 28 Jun 2014, 18:39:19 UTC 28 Jun 2014, 19:31:46 UTC Aborted by user 3,033.70 3,016.95 --- ACT-R cognitive modeling environment using Clozure Common Lisp (Linux Only) v2.32 (sse2) 17589299 16031680 28 Jun 2014, 18:40:47 UTC 28 Jun 2014, 19:31:46 UTC Aborted by user 2,957.94 2,939.31 --- ACT-R cognitive modeling environment using Clozure Common Lisp (Linux Only) v2.32 (sse2) 17589298 16031679 28 Jun 2014, 18:40:47 UTC 28 Jun 2014, 19:31:46 UTC Aborted by user 2,958.97 2,942.36 --- ACT-R cognitive modeling environment using Clozure Common Lisp (Linux Only) v2.32 (sse2) 17589288 16031669 28 Jun 2014, 18:38:43 UTC 28 Jun 2014, 19:31:46 UTC Aborted by user 3,081.67 3,059.96 --- ACT-R cognitive modeling environment using Clozure Common Lisp (Linux Only) v2.32 (sse2)

Profile Tom
Volunteer moderator
Avatar
Send message
Joined: 23 Jun 08
Posts: 490
Credit: 238,767
RAC: 0
Message 3235 - Posted: 8 Jul 2014, 13:46:24 UTC - in response to Message 3224.

Thanks Grandpa,

Our models are generated from many different researchers across many different projects, so runtimes can vary drastically between projects, jobs, and even individual work units (due to stochasticity inherent in the models). Furthermore, our project specializes in parallel parameter evaluations of cognitive models, meaning we run many simulations under varying conditions. We checkpoint workunits at the simulation level, and target workunits to take about an hour. If a single simulation takes 30-45 minutes, then we will run only one simulation per workunit. Unfortunately, in this case it means there will be no checkpointing, and no visible progress, until the workunit completes. Workunits would jump from 0% to 100% complete when the first simulation completes.

Our ideal and most efficient models take a couple minutes to run a single simulation. We then bundle enough of these simulations in one workunit to equal about an hour of runtime (to minimize bandwidth and overhead).

Let me know if that helps clarifies things.

Travis Weston
Send message
Joined: 18 May 14
Posts: 1
Credit: 6
RAC: 0
Message 3318 - Posted: 7 Nov 2014, 15:22:45 UTC

Any word on when work will be restarted?

Profile Conan
Avatar
Send message
Joined: 24 Mar 09
Posts: 60
Credit: 1,039,032
RAC: 106
Message 3319 - Posted: 7 Nov 2014, 23:29:37 UTC

Travis, there has been work available for a few days now, for all platforms.

Just allow work for your computers and you should get some.

Check the main page and it shows that there is a lot of work currently available.

Conan

Edvards .G
Send message
Joined: 1 Mar 14
Posts: 1
Credit: 5,687
RAC: 0
Message 3324 - Posted: 30 Nov 2014, 17:20:04 UTC - in response to Message 3319.

There is no more work available.. is there a database issue?

It's been like this for 2 days.

Dayle
Send message
Joined: 10 Oct 13
Posts: 9
Credit: 571,266
RAC: 85
Message 3325 - Posted: 15 Dec 2014, 19:41:20 UTC

Hopefully in the meantime you can create Windows versions of a few more of your applications. I'd love to be able to donate my processor time without switching my OS.

HighTech67
Send message
Joined: 9 Sep 08
Posts: 8
Credit: 1,247,807
RAC: 442
Message 3330 - Posted: 12 Jan 2015, 2:13:37 UTC - in response to Message 3325.

Hopefully in the meantime you can create Windows versions of a few more of your applications. I'd love to be able to donate my processor time without switching my OS.


Either that or start using VirtualBox.
____________

Message boards : News : Temporary Suspension of Work


Main page · Your account · Message boards


Copyright © 2017 MindModeling.org