Hi there,

Recently, I added the "Atlas" project to a PC that has plenty of memory to run VirtualBox.

On the same PC (running Ubuntu Linux with virtualbox 4.3.10-dfsg-1) I am running WCG and Einstein.

I put in an app_config.xml to limit the the number of Atlas that are running at once:
...
<max_concurrent>2</max_concurrent>
...

I see this error in the task status:
'Waiting to run (Scheduler wait: VM job unmanageable, restarting later)'

If I set all projects to "Won't get new tasks" ,
all running WU run to completion - EXCEPT the task mentioned above.

If I exit boinc-manager and restart the boinc-client,
Then the task completes.

Is this a bug or a feature?? :-)

I would like to see if I can get Atlas to play well with others.
I realize that it may take a day or two for BOINC to get some run-averages.

Meanwhile, I 'grew' the /var partition (with BOINC directory) from from 15 (or 20) GiB to 150GiB.
Afterwards, another WU hung with the same status. This time, I didn't wait, just suspended all WU; stopped and restarted the manager and client. The 'stuck' task restarted and finished OK.

I *believe* the error occurs when boinc does a task switch to another task taht was waiting.

Another user helpfully suggested that I set the "switch between applications" to 99999 minutes.
Yup. That works.

I did want to pass the error by you to see if anyone is interested.
Another user reported the same error in their team forum - but on the T4T task - which also uses VirtualBox.

[from the _Rolling Stone_]:
Suggestions?
Love-letters?
Advice?

added logging?

My gratitude and thanks,
Jay