Advanced search

Forums : Technical Support : Issues with vbox when working
Message board moderation

To post messages, you must log in.

AuthorMessage
icollins99

Send message
Joined: 19 Feb 20
Posts: 8
Credit: 6,358
RAC: 0
Message 22646 - Posted: 24 Sep 2020, 19:59:59 UTC

Hello!

For a while now, I have had some severe issues with working on any assigned tasks that require vbox. the main issue is that projects get assigned to my machine no problem, but once it reaches .100% progress, it stops all together, still claiming that it is working on the assignment, and continues to count up, with the estimated return time exceeding 10 days. I do have CPU virtualization enabled for my processor, 3600X, and the vbox manager lists each VM as working, and I'm able to access the machine. I have no actual experience with VMs, so any help would be greatly appreciated!
ID: 22646 · Report as offensive     Reply Quote
Jonathan

Send message
Joined: 27 Sep 17
Posts: 159
Credit: 7,580,022
RAC: 1,767
Message 22647 - Posted: 24 Sep 2020, 21:54:14 UTC - in response to Message 22646.  

Your tasks are all showing as camb_legacy and not camb_boinc2docker. If you want to run the VM versions (camb_boinc2docker) you need to select that in your cosmology@home preferences.

Max # CPUs controls how many cpu cores are assigned to a VM. I have mine set to 4 on a Ryzen 1700 that has eight true cores.
Max # of jobs controls the number of concurrent jobs, I think?

You will need to set no new tasks on the project, abort all current cosmology work units and then make your preference changes on this website. You can also check out the FAQ section. When you allow new work, you should get camb_boinc2docker work units with your new settings.
ID: 22647 · Report as offensive     Reply Quote
icollins99

Send message
Joined: 19 Feb 20
Posts: 8
Credit: 6,358
RAC: 0
Message 22648 - Posted: 26 Sep 2020, 1:46:09 UTC - in response to Message 22647.  

So, this is going to seem strange, but the tasks that are failing are the camb_boinc2docker tasks. The only ones that seem to be going through are the camb_legacy tasks. Shoulda put that in the post, huh? My CPU isn't too far off from yours, I'm using a Ryzen 3600X for all my tasks. I seemed to be having the same problem wen running tasks from LHC@home, whom also uses virtualization. I will definitely check my settings to see if anything there got bunged up, so fingers crossed!
ID: 22648 · Report as offensive     Reply Quote
icollins99

Send message
Joined: 19 Feb 20
Posts: 8
Credit: 6,358
RAC: 0
Message 22649 - Posted: 26 Sep 2020, 4:16:13 UTC - in response to Message 22648.  

Checked the settings, nothing there that seemed to be limiting it. I had my CPU and resources set to maximum, so I reset everything to unhinged again, to be sure. No luck there, though.
ID: 22649 · Report as offensive     Reply Quote
Jonathan

Send message
Joined: 27 Sep 17
Posts: 159
Credit: 7,580,022
RAC: 1,767
Message 22650 - Posted: 26 Sep 2020, 10:40:59 UTC - in response to Message 22649.  

Your computer doesn't show when I use the link from your forum name. LHC@home has a checklist for VM related task by Yeti. Try working through that to see if it helps. Do you have Hyper-V enabled? That clashes with Virtualbox but there may be a work around.
I am guessing you have your computer under a different account than the one you are posting on the forums under. Please post a link to your current computer so we can look at tasks and any error messages.

Link to computers showing on your forum account.
https://www.cosmologyathome.org/hosts_user.php?sort=rpc_time&rev=0&show_all=1&userid=1569122
ID: 22650 · Report as offensive     Reply Quote
icollins99

Send message
Joined: 19 Feb 20
Posts: 8
Credit: 6,358
RAC: 0
Message 22651 - Posted: 27 Sep 2020, 0:19:18 UTC - in response to Message 22650.  

My computer hasn't reported work in the past 30 days, so it doesn't show in the first part, but it is listed under "all computers." Checking my computer stats, it does say that hyper-v is enabled, and it is switched on in by BIOS. Looking at VBox, it lists al log the VMs that were created as inaccessible, with the error code -103, path not found, but has listed assigned tasks as working or paused in the past, so I'm guessing that a restart might have killed it. But, even in a situation like this, it seems to always return to normalcy, sans actually working, getting new assignments and opening new VMs without conflict. Here's my computer, hopefully you can see something I don't. https://www.cosmologyathome.org/show_host_detail.php?hostid=427527
ID: 22651 · Report as offensive     Reply Quote
Jonathan

Send message
Joined: 27 Sep 17
Posts: 159
Credit: 7,580,022
RAC: 1,767
Message 22652 - Posted: 27 Sep 2020, 2:23:13 UTC - in response to Message 22651.  

I am just double checking that you are using Microsoft Hyper-V and have it enabled or did you mean AMD-V or SVM is enabled in your BIOS?
Try setting your preferences to run just the camb_leagacy tasks and make sure those work. If you are running any other projects, stop or suspend them temporarily. If you can complete a few tasks then you can switch to running only the camb_boinc2docker version and work out the VM problem. Set max # cpus to six or less in your case or you can use the 'app_config.xml' method to control number of concurrent tasks running or camb_boinc2docker and the number of cpus cores assigned to a single VM. Just a wild guess right now is that the VM is getting 12 cores assigned and it is messing with all your work units. You can look at the settings of the running VM in Virtual Box
ID: 22652 · Report as offensive     Reply Quote
icollins99

Send message
Joined: 19 Feb 20
Posts: 8
Credit: 6,358
RAC: 0
Message 22653 - Posted: 27 Sep 2020, 2:49:38 UTC - in response to Message 22652.  

Sorry, my terminology is a bit rusty. I would assume SVM or AMD-V is on, as I am showing virtualization it enabled through task manager. I do not have Microsoft Hyper-V enabled, as far as I know, but I have also never seen the setting. I do have Windows defender core isolation off, although I don't think it would have much to do with my problem. I did see, when I looked up AMD-V, It said something about IOMMU, and how those settings affect GPU pass-through, but I don't know how that would affect me in my case. I have all of my settings changed how you said, and I'm waiting for it to chew through all of my legacy tasks, which thankfully take no time at all, so it can run docker in peace. I checked the settings for the VM, and it lists only 9 CPU cores assigned to it, which is what was listed in the description of the task. The rest of my tasks, which run alongside Cosmology@Home, don't seem to have throttling issues, as they all finish before deadlines.
ID: 22653 · Report as offensive     Reply Quote
icollins99

Send message
Joined: 19 Feb 20
Posts: 8
Credit: 6,358
RAC: 0
Message 22654 - Posted: 27 Sep 2020, 2:56:58 UTC - in response to Message 22653.  

Alright, so a few minutes after letting it run, the test failed, stating computational error, closing the VM. When I had last checked it, the estimated arrival was showing 12 days, with it stuck at .1% completion, which I would guess is the lowest number to register. I'll post the properties panel. [img]https://ibb.co/mXtpJQF[img]
ID: 22654 · Report as offensive     Reply Quote
icollins99

Send message
Joined: 19 Feb 20
Posts: 8
Credit: 6,358
RAC: 0
Message 22655 - Posted: 28 Sep 2020, 0:30:02 UTC

So, I was able to find the logs spit out by the most recent VM, so hopefully it will be able to reveal something. It did see that there was an error in the settings for the VM, saying that the number of CPUs exceeded the number available (paraphrasing, I forgot to copy the warning) but wouldn't allow me to change the number of CPU's I wanted the VM to use. I did have my settings set to only 6 CPUs through Cosmology@home after I began this thread, so fingers crossed that with new assignments it is able to assign the proper number of cores.

[url]http://www.mediafire.com/file/mz6c3ryet25d55r/boinc_f2f2a25a735bfeb9-2020-09-27-18-18-01.log/file[url]
ID: 22655 · Report as offensive     Reply Quote
Jonathan

Send message
Joined: 27 Sep 17
Posts: 159
Credit: 7,580,022
RAC: 1,767
Message 22656 - Posted: 28 Sep 2020, 0:46:37 UTC - in response to Message 22654.  

Your computer listing at Cosmology@home should be listing current assigned tasks. I want to try to get you new tasks and let you try an 'app_config.xml' set up to control the camb_boinc2docker application.

Set no new tasks for the Cosmology project in the Boinc Manager and then abort all related tasks for this project.

Open notepad and paste the following text into it and do a 'save as' using the name 'app_config.xml' and make sure to change the Save As Type drop down at the bottom of Notepad from 'Text Document' to 'All Files'. You will save it in the project data folder that may be currently hidden. Program Data is a hidden folder on Windows 10. My location is 'D:\ProgramData\BOINC\projects\www.cosmologyathome.org'. You can either choose to show hidden files in the file browser or just copy and past that location but change the drive letter to where you installed Boinc.

After doing all that, exit Boinc and all tasks, restart Boinc Manager and allow new tasks for this project. It should grab some work units and then try to run them. These should also allow your computer to show current assigned tasks. I set the configuration below to run a single camb_boinc2docker task that uses 6 cores. It should run pretty quick as I finish in 5.5 to 6 minutes on 4 cpus. You can always edit the numbers to allow more concurrent tasks or change the cpu counts but I don't think you can go higher than 6 in your case.

app_config.xml
<app_config>
    <app>
        <name>camb_boinc2docker</name>
        <max_concurrent>1</max_concurrent>
    </app>
    <app_version>
        <app_name>camb_boinc2docker</app_name>
        <plan_class>vbox64_mt</plan_class>
        <avg_ncpus>6</avg_ncpus>
    </app_version>
</app_config>
ID: 22656 · Report as offensive     Reply Quote
icollins99

Send message
Joined: 19 Feb 20
Posts: 8
Credit: 6,358
RAC: 0
Message 22657 - Posted: 28 Sep 2020, 2:15:30 UTC - in response to Message 22656.  
Last modified: 28 Sep 2020, 2:15:52 UTC

Alrighty!! Looks like that XML script did the thing! It does seem to only be running on 6 threads, which is a lot more than 0 for certain, haha. What I found odd is that, like you saw, my preferences were set to 6 CPUs, but each new task was calling for 10! I'll make sure to keep that script in the folder, if only to make sure that it keeps the correct config. Thanks for all your help, this has really been getting my goat for a while now! Well, back to working with a working machine! Anything else that I should make sure is configured correctly before I set it off?
ID: 22657 · Report as offensive     Reply Quote
Jonathan

Send message
Joined: 27 Sep 17
Posts: 159
Credit: 7,580,022
RAC: 1,767
Message 22660 - Posted: 1 Oct 2020, 4:54:43 UTC - in response to Message 22657.  

I think you should be fine for now. I still don't see any tasks listed by your computer in the past 30 days. Are they completing okay? If you are running LHC@home VM projects, you will have to do the same thing for those. I usually only run one VM related task at a time.
ID: 22660 · Report as offensive     Reply Quote

Forums : Technical Support : Issues with vbox when working