Long running work unit

[JeromeC]

Good news, thanks for the answer.

But so does this mean that this WU is supposed to run more than 80 days ???

[Fritzr]

WU 6330999 is still running, but the tasks list shows it as "timed out...no response" today, a second WU also still running autoextended correctly.

Can this be reactivated or do I just toss the 53 hours it has run so far and abort it?

Suspending the task for now, so at least no more time will be wasted if it can't be reactivated.

[ChristianB]

@Fritzr: It can be activated but the better question to ask is why it wasn't autoextended. Can you verify that the two tasks currently on this computer are running concurrently (at the same time).

I activated the task again and will try to monitor the auto extension.

[Fritzr]

They do run concurrently at times with both in "running" state. Often though, one is "Waiting to run" while the other is "running" and three tasks from other projects also run.

If there is a time window within which the autoextend must run, then that is the likely problem with the task failing to autoextend due to the BOINC manager not setting it to "running" status within that window. The full status message indicates that this may be what happened.

Status is "Waiting to run (Scheduler wait: VM Hypervisor failed to enter an online state in a timely fashion)"

[ChristianB]

The autoextensions must occur within the last 14 days of the deadline. During this time the task must run for 4h to send a message to the server that it is still alive.

As far as I know the "Scheduler wait: VM Hypervisor failed to enter an online state in a timely fashion" message can only be cleared by restarting the host.

[Fritzr]

Reinstalling Boinc resets the status message until the task attempts to run. It stops after 3 to 4 minutes with the Hypervisor failed to start message

Restarting the computer resets the status message until the task again attempts to run. It stops after 3 to 4 minutes with the Hypervisor failed to start message

The second task will start at the same time and has now recorded an additional 30+ hours from the time I first noticed the problem. Not sure if it is a problem with this computer (Boinc has to be reinstalled after a reboot because it cannot connect to localhost until reinstalled) or if it is the individual task. Both tasks appear to be the same version of the WU code though.

Oh well, this looks like it will remain a mystery for now.

[ChristianB]

The task 14925719 has still not send an alive message. Your wingman on this job finished this task in 2.5 hours. If you already have over 50 hours logged you should abort this task. You will get partial credit. You should also try to find the reason for not connecting so you don't have to reinstall BOINC everytime you boot. This looks to me like a permission problem. Did you try the latest 7.2.42?

[JeromeC]

This WU has been running for 35 days (ie CPU time) now but the % is always the same, 18%, so the remaining estimate is increasing all the time.

Is this normal ? the WU is eating CPU, the console says



Is everything under control ? how long will it take ?

Thanks.

[ChristianB]

Its not normal that it stays at 18%. Can you try to restart the client and see if this resolves this? If not please abort.

[JeromeC]

The client and/or computer has been restarted a number of times in those 31 days... Ain't there no way to check actual activity ? No log in the VM or somewhere ? I just have to send to the bin those 31 days ? (And what about credits in this case ?)

[ChristianB]

No, there is no additional check for activity. The percentage has to increase with every second or the control script is not working anymore. But you will get Credit even if the task is aborted.

[JeromeC]

Ok I (sadly) understand, I confirm that the % is fixed to 18,360%.

I have abandoned the WU after 857 hours of calculation



(or 594 ? or 706 ?)

Maybe it is related with this problem ?