open Beta Test for new cmsearch VM application

[Mouse King]

Hi. I notice the following: now all my WU stopped with ?error while computing? after virtual machines failed creating normal snapshot. Usually problem occurred after power off VM, when it is creating snapshot. After this next start VM cannot deleted previous snapshot and after some time it is stopped with ?error while computing?. No problem with operation memory or something else - this error of snapshot manage. I thought this occurred only when creating snapshot and power off. But this WU was stopped when snapshot was not creating (after 5 min) and can't start normally. After some time ?error while computing? occurred. However next WU also can't start normally (no progress and use CPU and memory, I reattached project then).

Here are my thoughts - in the past process of computing was more stable, now I don't belief that I can computing at least one WU. May be it will be good to give users who can manage VM personally a possibility to receive WU as it was before?

Or maybe it will be good to have a possibility to save two snapshots. This will give a possibility to return to one that was saved before problem occurred.

Besides I need to write that one of already commutated WU not replicated. It will be great if it can be resend to another cruncher.

[ChristianB]

The current problems with vboxwrapper are taken care of and we will update to a new version on Friday. There is still no fix for some problems you mentioned because the neccessary change inside vboxwrapper introduced more problems.

As for the not replicated WU you mentioned. This particular one is from a Batch that is faulty and will not be replicated. As soon as I find time again (exams coming up) I will grant credits to those who completed such faulty tasks and abort the Jobs.

[Coleslaw]

Sorry if this has been asked before (too many posts to go through), but does RNA have any way of limiting how many instances are run per system? I'm capable of running one instance since they can take upwards of 4GB at times on some of my systems, but those systems can't go beyond a single task. Some of my team members are getting ready to hop on board and they have that concern as well. Thanks in advance.

[ChristianB]

With the new VM application it is guaranteed that the used RAM never exceeds 4GB (at the moment). When using a newer BOINC Client (7.2.33) it is also assured that only so much VM tasks are started at the same time as the RAM settings allow. We will lower the max. RAM usage to 3 GB when I update the application on Friday. Usually the application takes up to 600MB of RAM when running (but allocates the remaining 3.4GB).

[Coleslaw]

With the new VM application it is guaranteed that the used RAM never exceeds 4GB (at the moment). When using a newer BOINC Client (7.2.33) it is also assured that only so much VM tasks are started at the same time as the RAM settings allow. We will lower the max. RAM usage to 3 GB when I update the application on Friday. Usually the application takes up to 600MB of RAM when running (but allocates the remaining 3.4GB).

Thank you for the fast response.

[ChristianB]

I updated the vboxwrapper to the latest version and decreased the memory limit to 3GB for Windows.

There will be no new work. The new version will be send out when old Tasks fail and a new instance of a Job is created. This could take some time. If you have unstarted RNA World tasks please abort them.

[Bikermatt]

Hello, This task has been at 98.765% for several days now. Should I abort it or let it run?

[ChristianB]

Hello, This task has been at 98.765% for several days now. Should I abort it or let it run?

No, please keep it running.

[melban]

OS: Windows 7
Ram: 32 GB
Work Unit: 6330944
Task: 14923526
Computer: 26463
Virtual Box Version: 4.2.16.x

Just a suggestion/observation I though I would mention.

I've lost about 120 hours of processing due to snapshots not being taken and having to restart my computer. I assumed that when I closed boinc and it shut down processes that it would have saved the machine state so it could resume later. I've been taking manual snapshots so i don't lose the progress. Shown will be two images. one of my boinc Virtual Box snapshot history. Another is the Test4Theory Virtual Box application which takes snapshots and deletes the history very frequently.

[ChristianB]

Please don't take snapshots yourself. If the vboxwrapper isn't doing this on it's own we have to fix this. This is the first post I see about the issue of not snapshoting. On all other logs I've seen the snapshots are working fine. I've also seen some logs where an old snapshot couldn't be deleted and clogged the slot directory until the task errored out.

What is the content of your stderr.txt?

[melban]

Here the link to the log. Would you suggest I continue to take manual snapshots since i've already started, stop taking snapshots, or delete old snapshots so they can merge into 1 current snapshot ?
http://melban.me/tmp/RNAWorld.2014.01.31.stderr.txt

[ChristianB]

From the log it seems that the vboxwrapper is checkpointing correctly. You should delete your manual snapshots, leave only the last one and see if the wrapper continues to make snapshots on its own. There is a possibility you can't see this in vboxmanager. The vboxwrapper version in v1.10 is more advanced than in T4T.