open Beta Test for new cmsearch VM application

[Beorn]

Good evening, zombie67! (actually it's very early morning for me )

From what I've learned so far: All VM workunits start with a relatively short deadline of 3 weeks. Unlike the XXL tasks which have to be extended manually, there is a trickle mechanism and a server-sided script or program which should extend the deadline, provided the host is still on it and online (at least from time to time).

However, the server-sided script seems not to work in all cases until now. That is why ChristianB has manually extended today all VM tasks still running (I guess it was an extension of one month, as far as I can see).

The BOINC manager does not 'know' if due date has been changed on the server or not. I assume that is the same case with the XXL tasks (never did those)?

Hope that helps (and I got it right).

[JeromeC]

Problem is I cannot be constantly checking how long is remaining / ask for WU deadline extension... I just lost 200 hours again with a yoyo ecm because I couldn't find again the topic on yoyo forum where to ask for deadline extension, it seems the WU immediately fell into error when deadline was reached...

I know beta testing is all about testing and implies risk, but for such long running WUs it's really a pity to know the end user can "lose" so many crunching hours (more than 400 hours for me)... so I'm not very keen on trying with those beta again.

PLUS I have mentioned in this same thread (read my posts above) a verified issue with your VM that actually causes a kernel panic on my mac if I run a windows 7 VM with Parallel Desktop - verified after contacting PD support and being directed by them on some incompatibility between VirtualBox and Parallel Destkop VMs being ran at the same time, knowing that I can run a T4T boinc VM + the PD win7 VM without any problem, it really happens when running your RNA boinc beta VM...

[ChristianB]

Problem is I cannot be constantly checking how long is remaining / ask for WU deadline extension... I just lost 200 hours again with a yoyo ecm because I couldn't find again the topic on yoyo forum where to ask for deadline extension, it seems the WU immediately fell into error when deadline was reached...

I know beta testing is all about testing and implies risk, but for such long running WUs it's really a pity to know the end user can "lose" so many crunching hours (more than 400 hours for me)... so I'm not very keen on trying with those beta again.

The current VM approach should eliminate the need to check for deadline extensions. Unfortunately there still seems to be an issue. This was fixed in vboxwrapper yesterday and I will deploy a new app soon where automatic deadline extension is working.

PLUS I have mentioned in this same thread (read my posts above) a verified issue with your VM that actually causes a kernel panic on my mac if I run a windows 7 VM with Parallel Desktop - verified after contacting PD support and being directed by them on some incompatibility between VirtualBox and Parallel Destkop VMs being ran at the same time, knowing that I can run a T4T boinc VM + the PD win7 VM without any problem, it really happens when running your RNA boinc beta VM...

I don't know how T4T created their VM but someone on the T4T forum said that they have disabled the "Enable VT-x/AMD-V" setting which we have enabled. This would indeed be problematic when running two visualization hypervisors. So I would conclude that you should disable the cmsearch VM app and also keep an eye on new T4T applications as they could introduce a new VM with "Enable VT-x/AMD-V" enabled.

[Beorn]

http://www.rnaworld.de/rnaworld/result. ... d=14920843

(..) Hope my wingman will make it back to the airfield now.

Yeah! He made it. Cheers to LCB001, Hardware Canucks, Canada! 6,078.56 credits, that's 26.08 cr. / hr. for me, however, less cr. / hr. for my wingman.

[ChristianB]

I just created a new set of Testjobs. The estimated runtime is circa 10h (with some exceptions). Please try to use your computer as usual and see if the tasks survive a restart. A lot of the former problems should be solved now. There should be only one running RNAWorld VM at the moment, this is a server setting.

[Crystal Pellet]

I just created a new set of Testjobs. The estimated runtime is circa 10h (with some exceptions). Please try to use your computer as usual and see if the tasks survive a restart. A lot of the former problems should be solved now. There should be only one running RNAWorld VM at the moment, this is a server setting.

RNA World 11 Nov 15:40:00 No tasks are available for cmsearch VM (VirtualBox) 1.0.2 ??

Server status: cmsearch VM (VirtualBox) 1.0.2 28 unsent

Could be the reason that I've installed VirtualBox 4.3.2 on that machine?

[Michael H.W. Weber]

Could be the reason that I've installed VirtualBox 4.3.2 on that machine?

Yes. You need to follow our installation instructions as detailed on the RNA World mainpage if you like to participate in the testing efforts.

Michael.

[Crystal Pellet]

Could be the reason that I've installed VirtualBox 4.3.2 on that machine?

Yes. You need to follow our installation instructions as detailed on the RNA World mainpage if you like to participate in the testing efforts.

Michael.

OK, no RNA then on that machine.

To test the newest vboxwrappers, I want/have to use also the newest BOINC 7.2.28 and the newest VirtualBox to be sure we will not get surprises, when everyone updates in future.
I loaded 1 task on my older machine, where also an almost 8 weeks Rattus-norvegicus is running; now in a stanalone cage.

[Jacob Klein]

Could be the reason that I've installed VirtualBox 4.3.2 on that machine?

Yes. You need to follow our installation instructions as detailed on the RNA World mainpage if you like to participate in the testing efforts.

Michael.

Are you blacklisting 4.3.2? If so, why?

I am aware of a 4.3.2-specific bug that I reported here:
https://www.virtualbox.org/ticket/12291
... but I thought it didn't affect RNA World.

As such, I would think that your project should now support VirtualBox 4.3.2, and support the users that have that version.
Could you please explain?

Thanks,
Jacob

[ChristianB]

I simply forgot to change the plan_class spec and lift the restriction when I updated the vboxwrapper yesterday. I fixed that now. There should be no project level restriction now only what the vboxwrapper blocks (4.2.6,4.2.18,4.3.0).

[Michael H.W. Weber]

I still did not get it, Christian.
To my knowledge, the latest BOINC version is fine (I am using it successfully myself), but all VirtualBox versions except 4.2.16 carry a bug which led us to block machines having installed one of these buggy versions from getting WUs, right? Please specify once again.

Michael.

[Jacob Klein]

I still did not get it, Christian.
To my knowledge, the latest BOINC version is fine (I am using it successfully myself), but all VirtualBox versions except 4.2.16 carry a bug which led us to block machines having installed one of these buggy versions from getting WUs, right? Please specify once again.

Michael.

VirtualBox 4.2.18, 4.3.0, and 4.3.2, all 3 have different bugs, and the first 2 bugs are fixed.
The remaining bug in 4.3.2 does not affect RNA World, since RNA World uses the "Enable VT-x/AMD-v" option.
Please see the following for details:
http://lhcathome2.cern.ch/test4theory/f ... 5560#15560

Test4Theory is still waiting for this 4.3.2 bug to be fixed, before they can recommend using 4.3.2:
https://www.virtualbox.org/ticket/12291
Again, my testing indicates that RNA World is not affected by this bug.