VboxHeadless running without vboxwrapper on Mac OS

[ChristianB]

That's not good. Because without the wrapper therre are no snapshots anymore. Can you please try to restart the BOINC client and see if the RNA VM is managed again? According to Rom this should work.

[JeromeC]

Basically one or the other VM project (RNA/T4T) are having the issue and then working again after some time / some event (unknown).

If I restart everything it works "for a while" and then the problems happens again, if I have more than one VM project.

Yesterday T4T was in this state but also full of I/O errors inside of the VM (stderr showed me that) so I canceled if and kill the VM.

Unfortunately I have family at home since last week-end staying in the room where the Mac is so it is hibernated in the night and not always on in the day, I won't be able to do any serious testing in the forthcoming weeks.

[JeromeC]

Since early july what happened is that it was always the T4T (now vLHC) WU that would experience the issue, the RNA WU was running high priority. I have extensive emails exchange with boinc dev team how provided me with test boinc version to help track down the issue (mainly Rom and Charlie, with better error reporting from the VM and the client) but unfortunately the issue was continuing, but then it would work, and so on.

But now I had to cancel the RNA WU (re the long WU topic here), so I don't know what happen, who knows when I get a new RNA, we'll see.

[ChristianB]

I saw the mails from BOINC Dev too and appreciate your effort in this matter.

[JeromeC]

Hello

I didn't have anymore input for that email thread with boinc devs, although I had sent updates of my testings.

I had frozen RNA "waiting for better times", some time ago I released it but hadn't got any new WU for a long time.

Also I did a setup (app_config) to let only one vLHC WU and one Atlas run at a time (especially since they allowed 2 vLHC running at the same time, it was not possible before), and things looked like OK between those two project applications (but I didn't have a lot of Atlas anyway, but it looked they could run together, again).

Yesterday evening when I came home I had got a new RNA WU, by that time I had a vLHC running and another one waiting to start the the 1st one would finish (re my app_config that prevents double execution). All looked OK then, I could check the console and see activity for woth WUs VM (RNA and vLHC), both were running and eating CPU in memory, both wrapper could be seen among boinc active processes.

But in the morning, that same situation had happened again : the two vLHC were both in the "pending unmanageable" state (that happens when the wrapper got lost in space), none of these were present in memory, no more wrapper for them, apparently RNA was still running OK and I didn't have time to check further. I only decided to cancel those two dead vLHC WUs and freeze vLHC to let RNA a chance for a while.

Now I just came home and I see the following : RNA "looks like it's running", the VM WU is eating CPU in memory, there is a visible wrapper also, but it takes ages for the console to show something (but finally it did show something saying "worker process with PID"), however the stderr.txt of the RNA task has not been updated since 00:17 tonight, and last entry in it is

2014-10-31 00:05:49 (23260): Status Report: Elapsed Time: '12000.503043'
2014-10-31 00:05:50 (23260): Status Report: CPU Time: '10920.110000'
2014-10-31 00:17:11 (23260): Creating new snapshot for VM.
2014-10-31 00:17:11 (23260): Restoring VM Process priority.
2014-10-31 00:17:24 (23260): Lowering VM Process priority.
2014-10-31 00:17:24 (23260): Deleting stale snapshot.
2014-10-31 00:17:31 (23260): Checkpoint completed.

meaning it didn't take any snapshot since almost 21 hours...

So what does it mean ? is this WU running or is it "dead" ? It says "remaining estimate 277 hours"... I will not let it run for a long time if it's not worth the wait.

Thanks for your advice.

[ChristianB]

Quick guess would be that you need a restart to really say if it is dead. But it's strange that the vboxwrapper is running but not logging anything.

[JeromeC]

Well... the situation was the same this evening, the advance % was stuck on 0,46%. So I did stop and restart boinc, the WU said it was running again, the wrapper was loaded into memory, but not the VM, no vboxheadless process anywhere, one core is not running at all.

The std_err was updated :

2014-10-31 00:17:31 (23260): Checkpoint completed.
<nothing between these 2 rows>
2014-11-03 22:52:16 (82164): vboxwrapper (7.3.26086): starting
2014-11-03 22:52:16 (82164): Feature: Checkpoint interval offset (476 seconds)
2014-11-03 22:52:16 (82164): Feature: Enabling trickle-ups (Interval: 14400.000000)
2014-11-03 22:52:17 (82164): Detected: VirtualBox 4.3.12r93733
2014-11-03 22:52:17 (82164): Detected: Sandbox Configuration Enabled

but it's been 10mn since 22:52 and nothing new has happened, the std_err was not updated again and the VM didn't show up... this WU was definitively dead and I let it 4 days eating a full core for nothing.

I decided to cancel the zombi WU.

I'll be back in a few months to see how things go.