Scheduler Wait (VM job unmanageable, restarting later).

[ChristianB]

I don't know. But there was a problem with El Capitan and vboxwrapper but I don't remmember the details. The wrapper will retry after 24h (max) if it is still not working abort the task.

[ChristianB]

I updated the vboxwrapper to the latest version for all platforms. You can try to get another task and it should use version 1.16 of the cmsearch VM app (which has the new wrapper).

[BobCat13]

Got another task last night using the newest vboxwrapper, but it didn't go well.

The first attempt was with VBox 4.3.30 installed and after 2.50 hours it went into the unmanageable state due to not deleting the old snapshot. I stopped the client, removed the old snapshot using VBox gui and started the client again. One hour into that run, same thing happened. Stopped the client again, downloaded and installed VBox 5.0.14 and edited the client state file and cleaned the slot directory to make the task start from the beginning. One hour of running, and unmanageable due to not deleting the old snapshot, stopped the client, removed the old snapshot and restarted the client. Another half hour and same thing happened, so I aborted the task.

Linux Mint 17.2, VBox 5.0.14 with extensions installed.

[ChristianB]

Hi BobCat13,

do you still have the logfiles of that tests? I wonder if this is the same issue I heard from another project where hey can also reproduce it.

[Michael H.W. Weber]

In the ATLAS@home forum a guy said that older BOINC versions cause trouble with VirtualBox snapshots in the sense that there is some sort of occasional I/O delay which results in improperly written snapshots that cannot be resumed. I am not sure whether that is correct but I do indeed have some problems with ATLAS@home on a machine which has an older BOINC version - the error rate is statistically significantly higher compared to other machines with an updated BOINC. So you also might want to check your BOINC version (most Linux distributions usually provide only outdated BOINC clients in their software repositories).

Michael.

[BobCat13]

Hi BobCat13,

do you still have the logfiles of that tests? I wonder if this is the same issue I heard from another project where hey can also reproduce it.

Yes, my daily rsync data from yesterday was still there as it has not run yet today, but it only includes the last test as I started the task from the beginning for that try. Which files do you want and where to send them?

The boinc core client is ver. 7.4.41 in case that helps.

[ChristianB]

Can you please send the vbox logs and the stderr.txt to christian.beer_a_rechenkraft.net ? Thanks

[BobCat13]

Files have been sent.

[ChristianB]

I got them thanks. I'm probably going to forward those to the vboxwrapper developer.

[JeromeC]

I have, on a Mac (latest OS X) with boinc 7.6.22 and VB 5.0.10 I tried RNA again after a very long stop with RNA (I have been able to run very few RNA in the past, especially since it went VM/VB), I see it runs 20mn and stops, I found this in the error log, repeated :

016-02-01 16:41:52 (56020): Creating new snapshot for VM.
2016-02-01 16:43:18 (56020): Deleting stale snapshot.
2016-02-01 16:43:32 (56020): Error in delete stale snapshot for VM: -2147467259
Command:
VBoxManage -q snapshot "boinc_b54425be7c6068a6" delete "6cb238d6-5a0b-4178-8850-69de6543c7a1"
Output:
0%...
Progress state: NS_ERROR_FAILURE
VBoxManage: error: Snapshot operation failed
VBoxManage: error: Code NS_ERROR_FAILURE (0x80004005) - Operation failed (extended info not available)
VBoxManage: error: Context: "RTEXITCODE handleSnapshot(HandlerArg*)" at line 532 of file VBoxManageSnapshot.cpp

2016-02-01 16:43:32 (56020): ERROR: Checkpoint maintenance failed, rescheduling task for a later time. (-2147467259)

I can send the full log (3 txt files in the slot dir).

The task is in state "VM unmanageable". Should I abort it ?


I just found out there's a newer VB (5.0.14) so I just installed it to see what happens, it can't be worse... can it ?

[MLx]

I'm getting the same error with VBox 5.0.14. (BOINC 7.6.22, OS X 10.11.4). Task link.

Just noticed there's a 5.0.16 update, will update and see if it helps.

[MLx]

Didn't help, still the same error:

Code: Alles auswählen

2016-03-29 12:01:49 (57526): Creating new snapshot for VM.
2016-03-29 12:01:54 (57526): Deleting stale snapshot.
2016-03-29 12:01:54 (57526): Error in delete stale snapshot for VM: -2147467259
Command:
VBoxManage -q snapshot "boinc_a1bd11c312417537" delete "572b5ecd-085d-415d-8617-fa7cb9e132a9" 
Output:
0%...
Progress state: NS_ERROR_FAILURE
VBoxManage: error: Snapshot operation failed
VBoxManage: error: Code NS_ERROR_FAILURE (0x80004005) - Operation failed (extended info not available)
VBoxManage: error: Context: "RTEXITCODE handleSnapshot(HandlerArg*)" at line 532 of file VBoxManageSnapshot.cpp

2016-03-29 12:01:54 (57526): ERROR: Checkpoint maintenance failed, rescheduling task for a later time. (-2147467259)
2016-03-29 12:01:54 (57526): Powering off VM.
2016-03-29 12:01:54 (57526): Successfully stopped VM.