VirtualBox

Ticket #9604 (closed defect: fixed)

Opened 3 years ago

Last modified 20 months ago

Restoring snapshot fails, crashes all running instances => Fixed in SVN

Reported by: mkns Owned by:
Priority: major Component: other
Version: VirtualBox 4.1.2 Keywords: snapshot
Cc: Guest type: Linux
Host type: Linux

Description (last modified by Hachiman) (diff)

Hi,

I was running the 4.1 version of Virtualbox on a Debian Squeeze, 2.6.32-5-amd64 kernel machine.

I had seen twice an attempt to restore a snapshot using VBoxManage fail:

user@host~$ VBoxManage snapshot virtualhost restore post Restoring snapshot 65ec2064-0601-49e3-ac0c-46b2ebc66f6b 0%...10%...20%...30%...40%...50%... Progress object failure: NS_ERROR_CALL_FAILED VBoxManage: error: Snapshot operation failed. No error message available!

What was even more concerning, however, was that all other running virtual instances immediately crash, such that VBoxManage list runningvms returns zero instances. This is a bit of a killer problem, as I was going to run this on a server for multiple people each of who would have multiple instances, but the thought of one person trying to do a restore and crashing every other instance is clearly not appetizing.

I raised this here, but got no response:

 http://forums.virtualbox.org/viewtopic.php?f=7&t=43951

So, I decided that maybe Virtualbox didn't like a 64-bit OS, so today re-installed the machine entirely with Debian Squeeze 32-bit. Alas, the exact same problem happened, and this time I was using the GUI to take and restore the snapshot. Taking the snapshot works, restoring the snapshot crashes everything.

Further information from dmesg isn't particularly helpful, to me, anyway:

[ 8646.862243] VBoxSVC[12389]: segfault at 0 ip 0814f102 sp b63140f0 error 4 in VBoxSVC[8048000+21a000]

I can only guess therefore that Virtualbox doesn't like the hardware, since it's a total re-install. This machine is pretty powerful; two Intel X5650 processors, 32GB or RAM, 15k disks.

If anyone can suggest what to look into, I'd appreciate it, because Virtualbox is totally awesome except for this snapshot issue. Please help me move my team away from VMWare Server 2!

Change History

comment:1 Changed 3 years ago by mfox

I'm receiving the same error running VB 4.1.2, Ubuntu 11.04 64-bit host with a AMD Phenom II. Snapshot creation works fine, but the restore causes the following:

{VBoxManage snapshot XXXXXX restore "XXXXXXXXX" Restoring snapshot 3e400449-21cd-4cb5-8db0-e0e62aa949b1 0%...10%...20%...30%...40%...50%... Progress object failure: NS_ERROR_CALL_FAILED VBoxManage: error: Snapshot operation failed. No error message available!}

The guest is Windows7 64 bit, but I also had the issue with Win7 32 bit. I've seen the problem on an Ubuntu 11.04 64bit guest as well. If you try to restore again it usually works. The issue seems random, but not VM specific. I dont see anything in the VM logs, so if there is any other data that you need please let me know.

I think this needs more attention because we're not talking about just crashing one VM, it brings down all running VMs. I've already corrupted more then one production system due to this issue.

comment:2 Changed 3 years ago by frank

VBox 4.1.4 contains an important bugfix regarding restoring snapshots. However, I'm not sure if the fix affects your problem so I would appreciate if you could test if there is any difference with VBox 4.1.4.

comment:3 Changed 3 years ago by mkns

Hi,

Thanks - I've downloaded and installed 4.1.4 today and will try taking/restoring snapshots during this week to see whether it is able to reliably not crash everything. I'll get back to you in a few days with the results.

comment:4 Changed 3 years ago by mfox

I just tried restoring a machine and hit the bug again. It looks like exactly the same issue and all running VMs were "Powered Off" when I looked at the list after the crash.

comment:5 follow-up: ↓ 12 Changed 3 years ago by frank

mfox, in that case it sounds like the VBoxSVC daemon is crashing during this operation for some reason and it leaves the mess behind. Could you provide a core dump of the crashing VBoxSVC daemon?

comment:6 Changed 3 years ago by mfox

frank, I'll be honest I'm not the best with Linux. I've been trying to get a core dump for a couple of days but I cannot seem to get it to dump. I've read your wiki page on the topic and I've also looked up other instructions on doing this and I just got the crash to work, but no dump. Is there another document that describes this a little clearer?

For cross reference, here are the two pages I was going by: https://www.virtualbox.org/wiki/Core_dump  http://www.randombugs.com/linux/core-dumps-linux.html

comment:7 Changed 2 years ago by mkns

Hi,

I'm afraid 4.1.4 has made no difference to my problem - I just tried to take and restore the snapshot, and the restore failed, crashing all running virtual machines at the same time:

mkns@server:~$ VBoxManage snapshot vminstance restore foo1 Restoring snapshot 084016e1-8c8b-4485-ac1e-531658f7fadd 0%...10%...20%...30%...40%...50%... Progress object failure: NS_ERROR_CALL_FAILED VBoxManage: error: Snapshot operation failed. No error message available!

I'm afraid I will be unable to send you any form of core dump for security reasons. If anything was logged to a log file, on the other hand, that would have been no problem... except for the fact that the error itself notifies me that "No error message available!".

comment:8 follow-up: ↓ 9 Changed 2 years ago by agent-smith

Hi,

I have same problem in build 4.1.8. Trying to restore to a snapshot sometimes giveme a segfaul in VBoxSVC & crashes all VMs. I can see

VBoxSVC[18966]: segfault at 0 ip 00000000004c349a sp 00007f59e40edbb0 error 4 in VBoxSVC[400000+280000]

in dmesg. How can i make a core dump to provide it here?

comment:9 in reply to: ↑ 8 Changed 2 years ago by Hachiman

  • Description modified (diff)

Replying to agent-smith:

How can i make a core dump to provide it here?

Please look at Core Dump article for more details, except that in command line you need to start VBoxSVC instead VirtualBox after

# ulimit -c unlimited

comment:10 Changed 2 years ago by haberb

This has been affecting me more and more, everytime I upgrade it seems to get worse. I'll try create a core dump.

comment:11 Changed 21 months ago by haberb

I have managed to generate a core dump, but I don't think I can upload it here. Any recommendations on how to get it to you?

comment:12 in reply to: ↑ 5 ; follow-up: ↓ 13 Changed 21 months ago by haberb

Replying to frank:

mfox, in that case it sounds like the VBoxSVC daemon is crashing during this operation for some reason and it leaves the mess behind. Could you provide a core dump of the crashing VBoxSVC daemon?

frank, see last comment, I have a core dump but not sure how to get it to you. it's almost 4 megs compressed

comment:13 in reply to: ↑ 12 Changed 21 months ago by haberb

Replying to haberb: Nevermind, emailed it to frank per link above.

Replying to frank:

mfox, in that case it sounds like the VBoxSVC daemon is crashing during this operation for some reason and it leaves the mess behind. Could you provide a core dump of the crashing VBoxSVC daemon?

frank, see last comment, I have a core dump but not sure how to get it to you. it's almost 4 megs compressed

comment:14 follow-up: ↓ 15 Changed 21 months ago by frank

  • Summary changed from Restoring snapshot fails, crashes all running instances to Restoring snapshot fails, crashes all running instances => Fixed in SVN

Finally! Thanks for the core dump. It turned out that this is the same problem we just fixed in #10491. This bug will be fixed in the next maintenance release expected soon.

comment:15 in reply to: ↑ 14 Changed 21 months ago by haberb

Replying to frank:

Finally! Thanks for the core dump. It turned out that this is the same problem we just fixed in #10491. This bug will be fixed in the next maintenance release expected soon.

Thank you so much, this bug was a nightmare. Can't wait to test it

comment:16 Changed 20 months ago by frank

  • Status changed from new to closed
  • Resolution set to fixed

The fix is part of the 4.1.20 release.

Note: See TracTickets for help on using tickets.

www.oracle.com
ContactPrivacy policyTerms of Use