KVM Migration - 3.0 to 3.1

lweidig

Active Member
Oct 20, 2011
104
2
38
Sheboygan, WI
We upgraded one of our host systems to VE 3.1 which was just released. Prior to that we live migrated a few containers and a few KVM guests to another server which of course went smoothly with no issues. After upgrading the host, rebooting for new kernel we tried to migrate back the machines to balance load again. All of the containers migrated back to the upgraded system, however we get something similar to the following for all of the KVM migrations:

Aug 21 09:29:35 starting migration of VM 103 to node 'xxx-01' (xxx.xxx.xxx.xxx)
Aug 21 09:29:35 copying disk images
Aug 21 09:29:35 starting VM 103 on remote node 'xxx-01'
Aug 21 09:29:37 ERROR: online migrate failure - unable to detect remote migration port
Aug 21 09:29:37 aborting phase 2 - cleanup resources
Aug 21 09:29:37 migrate_cancel
Aug 21 09:29:38 ERROR: migration finished with problems (duration 00:00:03)
TASK ERROR: migration problems

Thoughts on how to resolve this or what might be the cause? The KVM machines are typically Windows servers in our environment.
 
Not sure this is the same issue for us as what you linked though the result seems the same. We were previously running all 2.6.32-22 kernels on all of our hosts and had no issues migrating around servers. In fact the remaining ones still running that version, that have not been upgraded this still work for live migrataion of KVM servers.
 
I did some tests and all our servers with new proxmox 3.1 are affected and not only the old one (see my further post).

Regards,
Oer
 
I experience the same problem
I cannot upgrade from 3.0 to 3.1 without rebooting all vms because live migration from 3.0 to 3.1 does not work
Any idea for a workaround so that i can migrate a cluster without rebooting the vms ?
 
Hi, i've the same problem : i've a 2 node cluster in HA mode. For older versions i upgrade secondary node and reboot, after : migration for all vm on second node -> upgrade of the first node. That seems doesn't work between 3.0 and 3.1 version. The 2 servers are identicals : 2xAMD Opteron 6376 16 core, 128 Go de RAM, RAID card LSI with Cachecade 2, 12x2To (RAID6) and 2x240 Go SSD Intel (for cachecade). any idea ? I need to update my cluster without rebooting the vms.

best regards,

--
Christophe Casalegno
 
you need to upgrade all hosts to 3.1, also these with running VMs. then you are able to live-migrate.
 
you need to upgrade all hosts to 3.1, also these with running VMs. then you are able to live-migrate.

I've do this (but don't reboot one because i can't stop vms..) :

When i try to migrate HA VM :

task started by HA resource agent
Aug 30 21:49:01 starting migration of VM 216 to node 'netgrid2' (xxx.xxx.xxx.xxx)
Aug 30 21:49:01 copying disk images
Aug 30 21:49:01 starting VM 216 on remote node 'netgrid2'
Aug 30 21:49:02 starting ssh migration tunnel
Aug 30 21:49:03 starting online/live migration on localhost:60001
Aug 30 21:49:03 migrate_set_speed: 8589934592
Aug 30 21:49:03 migrate_set_downtime: 0.1
Aug 30 21:49:05 ERROR: online migrate failure - aborting
Aug 30 21:49:05 aborting phase 2 - cleanup resources
Aug 30 21:49:05 migrate_cancel
 
I verified it is possible to upgrade a cluster from 3.0 to 3.1 with all machines running
Just upgrade all machines in the cluster to 3.1 while the vms are running
Live migration is then possible and we can reboot the cluster-members one by one.
 
I verified it is possible to upgrade a cluster from 3.0 to 3.1 with all machines running
Just upgrade all machines in the cluster to 3.1 while the vms are running
Live migration is then possible and we can reboot the cluster-members one by one.

Nope !
When you upgrade a 3.0 host you have to reboot so make it running 3.1.
So if you want to reboot you need to migrate VM to another machine. And if this machine is running 3.1 you have the error mentioned above !
 
I am quite worried that regularly when upgrading to a newer proxmox release I have to reboot all my VMs because migration from an older to a newer proxmox version doesn't work. It's a pain in the ass !
 
your statement is simply not true, in most cases you can live migrate to keep everything running.
 
Sorry, I was wrong.
After an update one of my cluster did not restart in 3.1 and stay in 3.0 and I had to reboot it. I though it will be the same for all, but I was wrong. My other nodes updated from 3.0 to 3.1 without need to be rebooted thus I can do the live migration.
Sorry again, please forgot my previous post.
 
Thing is, you may still very much WANT to shutdown (not restart) and start all your KVMs after a dist upgrade, if it contained a KVM or qemu update, otherwise your VMs wont benefit from the new features (if there are any in the release).
 

About

The Proxmox community has been around for many years and offers help and support for Proxmox VE, Proxmox Backup Server, and Proxmox Mail Gateway.
We think our community is one of the best thanks to people like you!

Get your subscription!

The Proxmox team works very hard to make sure you are running the best software and getting stable updates and security enhancements, as well as quick enterprise support. Tens of thousands of happy customers have a Proxmox subscription. Get yours easily in our online shop.

Buy now!