New application version - 1.37


Advanced search

Message boards : News : New application version - 1.37

1 · 2 · Next
Author Message
etalhod
Send message
Joined: 4 Dec 14
Posts: 56
Credit: 30,151
RAC: 0
    
Message 2550 - Posted: 24 Jun 2015, 9:49:46 UTC

Dear All,

In the next hour a new application version for all platforms will be released.

The updates in this version are:

--New vboxwrapper version.
--Changes in the vbox_job.xml

The image will remain the same as in the old version.

Your inputs are important,
Please let us know if you are having problems.

Thank you all!

Mogens Dam
Send message
Joined: 1 Jul 14
Posts: 26
Credit: 3,264,917
RAC: 0
    
Message 2552 - Posted: 24 Jun 2015, 14:27:45 UTC - in response to Message 2550.

Could that explain why suddenly the "Remaining time" suddenly increased from typically 3 hours for a new job to 47 hours.

Are the jobs of the same size as before or are they really much, much larger?

-Mogens

Profile Jim1348
Send message
Joined: 15 Nov 14
Posts: 111
Credit: 1,319,885
RAC: 0
    
Message 2554 - Posted: 24 Jun 2015, 17:11:18 UTC - in response to Message 2552.

The remaining times are OK for me on two Haswell machines (BOINC 7.6.2 x64), being about 1 hour 25 minutes, which is quite close to the actual value. You probably just have a start-up problem.

Adam Myatt
Send message
Joined: 12 Apr 15
Posts: 4
Credit: 242,697
RAC: 0
    
Message 2555 - Posted: 24 Jun 2015, 17:35:58 UTC - in response to Message 2552.

I seen about the same thing. My jobs went from under 3 hours to ~36 hours estimated. I have exited BOINC manager and restarted and still the same result.

rbpeake
Send message
Joined: 27 Jun 14
Posts: 86
Credit: 8,794,961
RAC: 0
    
Message 2556 - Posted: 24 Jun 2015, 18:28:24 UTC - in response to Message 2555.

Boinc has to re-learn the correct computing time, which is similar to the old units. It will adjust properly after a few units.

I seen about the same thing. My jobs went from under 3 hours to ~36 hours estimated. I have exited BOINC manager and restarted and still the same result.

Profile MAGIC Quantum Mechanic
Avatar
Send message
Joined: 4 Jul 14
Posts: 332
Credit: 485,372
RAC: 0
    
Message 2557 - Posted: 24 Jun 2015, 19:20:22 UTC

Just picked up 13 of the 1.37 tasks.......now after an hour or so my slow dsl will finish the d/l so I can start running them.

And see if the ones that had problems with 1.36 will be able to run these.
____________

David Duvall
Send message
Joined: 29 Jun 14
Posts: 15
Credit: 1,191,257
RAC: 0
    
Message 2558 - Posted: 24 Jun 2015, 23:57:00 UTC

This new version fixed my problem where I was having to manually clean up my virtual media that wasn't getting deleted. Thank you.

etalhod
Send message
Joined: 4 Dec 14
Posts: 56
Credit: 30,151
RAC: 0
    
Message 2559 - Posted: 25 Jun 2015, 8:55:55 UTC

Thank you All!

More details about this version:

We are using now the BOINC vboxwrapper version 26166.
The image is the same as before and the job is triggered by a cronjob.

Profile tullio
Send message
Joined: 27 Jun 14
Posts: 256
Credit: 289,886
RAC: 0
    
Message 2560 - Posted: 25 Jun 2015, 10:31:07 UTC

All is well on my Windows 8.1 PC.
Tullio

JSE
Send message
Joined: 14 Feb 15
Posts: 137
Credit: 9,505,329
RAC: 0
    
Message 2561 - Posted: 25 Jun 2015, 17:38:21 UTC

All 1.37 running really stable on my Ubuntu Linux machines... Haven't had them that stable before!

Profile Yeti
Avatar
Send message
Joined: 20 Jul 14
Posts: 699
Credit: 22,597,832
RAC: 0
    
Message 2564 - Posted: 25 Jun 2015, 18:04:16 UTC

So far, all is running fine here on Windows. Had only 2 Tasks 1.37 that finished after 10 minutes and a wingman could crunch them normal.

All other seem to be fine.

The duration of the jobs seems to be the same as with 1.36

By the way one question: I'm still with VB 4.3.12. Will a newer release be better with the wrapper 26166 ?

etalhod
Send message
Joined: 4 Dec 14
Posts: 56
Credit: 30,151
RAC: 0
    
Message 2573 - Posted: 26 Jun 2015, 14:24:14 UTC

Thank you for your inputs!

VB 4.3.12 is OK for that release.

Profile Harm Joris ten Napel
Send message
Joined: 27 Nov 14
Posts: 40
Credit: 1,696,115
RAC: 0
    
Message 2574 - Posted: 26 Jun 2015, 15:21:53 UTC - in response to Message 2550.

This seems to be an improvement as I don't see 'yellow' tasks in unmanageable state anymore where I had to restart the boinc-client service on xubuntu to get them in the green again.

Profile Jim1348
Send message
Joined: 15 Nov 14
Posts: 111
Credit: 1,319,885
RAC: 0
    
Message 2575 - Posted: 26 Jun 2015, 16:25:44 UTC - in response to Message 2574.

Same for me with Win7 64-bit. Also, it seems to use a bit less disk memory, as was stated before. I use a 12 GB write-cache for the BOINC data folder, which allows me to run three ATLAS and three WCG/CEP2 work units, along with supporting a couple of GPUs. It used to overflow a couple of GB from the cache onto the disk after a day of use, but now it does so much less or not at all.

The working memory for the applications seems to be the same as before though. Overall, it is a nice improvement.

Profile MAGIC Quantum Mechanic
Avatar
Send message
Joined: 4 Jul 14
Posts: 332
Credit: 485,372
RAC: 0
    
Message 2576 - Posted: 26 Jun 2015, 18:34:39 UTC

1.37 working with no problems on Win7 and 8.1

Jim
Send message
Joined: 19 Jul 14
Posts: 56
Credit: 218,448
RAC: 0
    
Message 2577 - Posted: 26 Jun 2015, 18:35:46 UTC

I am seeing the same run times as with with 1.35, but the stability is much improved. I have yet to see the dreaded message about unmanageable VMs.

Running iMac, OS 10.10.3

Jim

Phil
Avatar
Send message
Joined: 27 Jun 14
Posts: 39
Credit: 383,974
RAC: 0
    
Message 2578 - Posted: 26 Jun 2015, 22:52:55 UTC
Last modified: 26 Jun 2015, 23:01:22 UTC

Yep I've tried pausing/shutting down/rebooting the host machine and this one is very stable.
VBox Manager no longer complains about missing images.

I'm sure though that some folk would like a "peek" inside the VM to see whats going on; the "show graphics" button is live but alas apache inside the VM isnt running and the click just gives an error message.

Profile Harm Joris ten Napel
Send message
Joined: 27 Nov 14
Posts: 40
Credit: 1,696,115
RAC: 0
    
Message 2580 - Posted: 26 Jun 2015, 23:42:48 UTC - in response to Message 2578.

Yeah, while fear that CERN my produce world consuming black-holes is exaggerated, these VM images sure are in for the competition: please enhance the methods to inspect what's going on inside them or you may lose popular support!

Profile tullio
Send message
Joined: 27 Jun 14
Posts: 256
Credit: 289,886
RAC: 0
    
Message 2586 - Posted: 28 Jun 2015, 8:17:28 UTC
Last modified: 28 Jun 2015, 8:17:49 UTC

Am I wrong or ATLAS@home is no longer making snapshots? This appears also from a lesser RAM usage on my Windows 8.1 PC.
Tullio

Andrej Filipcic
Project administrator
Project developer
Project tester
Project scientist
Send message
Joined: 3 Jun 14
Posts: 41
Credit: 3,891,911
RAC: 0
    
Message 2587 - Posted: 28 Jun 2015, 11:02:48 UTC - in response to Message 2586.

True, the snapshots were causing too many issues, however, pausing the WU still works as expected. On the memory usage, each WU uses about 2GB of RSS, while it shows almost 4GB in total. While running several WUs in parallel, 2GB is shared it seems among them.

1 · 2 · Next

Message boards : News : New application version - 1.37