Jump to content


Photo

Urgent help: Live migration stuck on pending state

Started by Riccardo Tempesta , 16 February 2015 - 03:55 PM
11 replies to this topic

Best Answer Tobias Kreidl , 16 February 2015 - 04:20 PM

I doubt it -- it mostly will deal with re-establishing interconnection information among the various hosts.  I think I've only seen an issue once or twice ever with a xe-toolstack-restart. I'd runit on the host on which your VM currently resides that you are trying to migrate. Of course, it may not terminate the migration, either, as it's not really something that should interferece. If it's not causing any otehr issues, I'd just leave it alone and let it time out.

-=Tobias

Riccardo Tempesta Members

Riccardo Tempesta
  • 24 posts

Posted 16 February 2015 - 03:55 PM

Hi there,

I started live migration of 2VMs btween two hosts and then I had to cancel the task by clicking "cancel" on XenCenter.

 

After that I started live migration of another VM that regularly started migrating.

 

Now, after more than 4 hours they are all stuck and I cannot perform any operation on the other machines such as reboot.

I checked the progress during the last two hours and nothing changes.

 

What can I do? If I restart xapi, do I risk any VM failure?

Thank you in advantage.

 

 

[root@yukon /]# xe task-list|grep pending -A 5 -B 5

 
 
uuid ( RO)                : 40dada56-e903-598a-0a13-b20aa7819fcd
          name-label ( RO): Async.VM.migrate_send
    name-description ( RO):
              status ( RO): pending
            progress ( RO): 0.982
 
 
uuid ( RO)                : a7a43124-f333-6c57-b750-3766b7ffa005
          name-label ( RO): Async.SR.update
--
 
 
uuid ( RO)                : 06f37235-dfd3-c082-d28f-22897475dc1b
          name-label ( RO): Async.VM.migrate_send
    name-description ( RO):
              status ( RO): pending
            progress ( RO): 0.002
 
 
uuid ( RO)                : 3cac64bd-1d80-945c-0a3e-f2bccc40b65f
          name-label ( RO): Async.SR.update
--
 
 
uuid ( RO)                : afe74da5-4b02-baa2-824c-b7fe0334fe58
          name-label ( RO): Async.VM.hard_shutdown
    name-description ( RO):
              status ( RO): pending
            progress ( RO): 0.000
 
 
uuid ( RO)                : 117f0f01-0f93-59a6-398c-e03515711aa9
          name-label ( RO): Async.VM.migrate_send
    name-description ( RO):
              status ( RO): pending
            progress ( RO): 0.975
 
 
uuid ( RO)                : 15ee6757-2cb2-8cbc-d1fd-96644a2acc51
          name-label ( RO): Async.SR.update
 

 



Tobias Kreidl CTP Member

Tobias Kreidl
  • 17,931 posts

Posted 16 February 2015 - 04:08 PM

You could run "xe task-cancel uuid=(UUID-of-task)" for ech specific task involved and see if you can stop it that way. It should time out if nothing else within 24 hours if it does not complete on its own.



Riccardo Tempesta Members

Riccardo Tempesta
  • 24 posts

Posted 16 February 2015 - 04:11 PM

Hello Tobias,

thank you for your reply. I tried with task-cancel and tried with "force" option too, but nothing changes.

 

What if I restart xapi? Is it dangerous while a migration task is in progress?



Tobias Kreidl CTP Member

Tobias Kreidl
  • 17,931 posts

Posted 16 February 2015 - 04:20 PM

I doubt it -- it mostly will deal with re-establishing interconnection information among the various hosts.  I think I've only seen an issue once or twice ever with a xe-toolstack-restart. I'd runit on the host on which your VM currently resides that you are trying to migrate. Of course, it may not terminate the migration, either, as it's not really something that should interferece. If it's not causing any otehr issues, I'd just leave it alone and let it time out.

-=Tobias


Best Answer

Riccardo Tempesta Members

Riccardo Tempesta
  • 24 posts

Posted 16 February 2015 - 04:24 PM

Ok, fine, I'll try leaving it alone and wait for a timeout.

Thank you for your help.



Tobias Kreidl CTP Member

Tobias Kreidl
  • 17,931 posts

Posted 16 February 2015 - 04:26 PM

Sure thing, Riccardo. It may time out sooner than that. As long as nothing else is broken, I'd just leave it alone. Keep us posted!

-=Tobias



Tobias Kreidl CTP Member

Tobias Kreidl
  • 17,931 posts

Posted 16 February 2015 - 04:28 PM

P.S. Have you looked in XenCenter at the notifications window and tried to see if you could cancel the task that way? Which version of XenServer is this?



Riccardo Tempesta Members

Riccardo Tempesta
  • 24 posts

Posted 16 February 2015 - 04:31 PM

I'm running a XenServer 6.2 .

I tried clicking "cancel" in the XenCenter but nothing changes, it just disables the button after that.

The only "broken" thing is the fact that I have a pending shutdown in a VM that does not restart, but I can wait since it is not yet in production.



Tobias Kreidl CTP Member

Tobias Kreidl
  • 17,931 posts

Posted 16 February 2015 - 04:51 PM

OK, then just leave it and it should time out sooner or later. :)



Riccardo Tempesta Members
  • #10

Riccardo Tempesta
  • 24 posts

Posted 17 February 2015 - 08:20 AM

Well, some updates: As you predicted they timed out in less than 24 hours, they took about 12 hours.

After that I tried to perform some action such as restarting a couple of VMs but it did not work.

 

I finally restarted the toolstack with "xe-toolstack-restart" and all is fine now.

 

Thank you for your help Tobias.



Chad patterson Members
  • #11

Chad patterson
  • 2 posts

Posted 20 March 2017 - 10:42 PM

uuuuuuggggghh> do not restart toolstack dring migration! iT KILLD MY vmS. really poor advice!



Tobias Kreidl CTP Member
  • #12

Tobias Kreidl
  • 17,931 posts

Posted 21 March 2017 - 12:07 AM

Great to hear, Riccardo!