PVE Migration V1.3 --> V1.5

[ChristianAC]

Moin Moin

Ich bin grad bei einer Migration von V1.3 auf die aktuelle V1.4/5.

Zuerst sollen es nur zwei Test-VMs sein. Dies bedeteutet aber, daß ich
im Cluster einen Mix aus 1.4/5 und 1.3 fahre.

Wenn ich nun eine VM von 1.3 nach 1.4/5 verschiebe, dann klappt dies zwar aber
wird mit einer Fehlermeldung beendet.
Die VM lässt sich dann auf der 1.4/5 Seite auch nicht starten.
Wenn ich nun auf 1.4/5 eine neue VM anlege und darein das qcow2-File kopiere, dann
kommt die VM auch sauber hoch.

Nun hab ich versucht die alten Infos der verschobenen VM zu löschen. Dies wird
aber auch mit einem lapidaen failed quittiert.

Wo muss ich von Hand überall rein um die VM zu löschen ??

English Version:

I am just in a migration from V1.3 to V1.4/5

First just two TEST-VMs should migrate. That means that I have a mix of V1.3 and V1.4/5 in the cluster.
When I move a VM from V1.3 to V1.4/5 the data are moved but it ends with an error.

This VM is anuable to start an the V1.4/5 side.

I can create a new VM an V1.4/5 and copy the qcow2-file into it. Then the VM starts up and is working.

Now I tried to destroy the moved VM but this also fails.

Where do I have to edit manually to destroy the old VM, or better where do I have to change the configuration
manually to make a V1.3-migrate VM to work.

Regards

Christian

 

[dietmar]

Sorry, please can you re-post your question in english?

 

[ChristianAC]

done

 

[dietmar]

First just two TEST-VMs should migrate. That means that I have a mix of V1.3 and V1.4/5 in the cluster.

mixing different versions is a bad idea.

When I move a VM from V1.3 to V1.4/5 the data are moved but it ends with an error.

what error?

This VM is anuable to start an the V1.4/5 side.

Why? What error do you get?

Now I tried to destroy the moved VM but this also fails.

Why? What error do you get?

 

[ChristianAC]

mixing different versions is a bad idea.

I don't have another chance. First upgrade one machine, moving VMs
and then upgrading the second.

what error?

Hmmm can reproduce it. Might have been a glitch. Migration of the second
VM without error.

Why? What error do you get?

Dont know. just failed.

Dec 30 13:25:26 polyphem pvedaemon[4447]: starting VM 103 on node 1 (localhost)
Dec 30 13:25:26 polyphem pvedaemon[4447]: VM 103 start failed -

Where to find detailed error-report ??

Why? What error do you get?

Also....

Dec 30 12:31:39 polyphem pvedaemon[2822]: destroying VM 105 on node 1 (localhost)
Dec 30 12:31:39 polyphem pvedaemon[2822]: VM 105 destroy failed -

Detailed report? Where?

Regards

Christian

 

[tom]

Logs:

start, stop, destroy, migrate... logs - see web interface:

Administration/Logs/Tasks

 

[ChristianAC]

Hmmmm

/usr/sbin/qm start 103
VM is locked (migrate)
VM 103 start failed -

stopped qemu-server

removed all under /var/lock

started qemu-server

start VM 103

same result.

Regards

Christian

 

[ChristianAC]

OK.

qm unlock 103

solved the problem.

problem of missed communication between V1.3 and V1.4/5 ??

regards

Christian

 

[tom]

OK.

qm unlock 103

solved the problem.

problem of missed communication between V1.3 and V1.4/5 ??

regards

Christian

thanks for update. a cluster with different versions is not supported.

 

[ChristianAC]

@Tom

But you should think of that!

When you update a HACMP-Cluster in an productiv environment you start with
updating the standby-node, moving all resources to the standby node and then
updating the primary node.
After that you move back the resources based on the policy. (cascading / rotating)

Normal bussiness.

Normaly you wont get the downtime from the customer to upgrade both nodes at once.

Regards

Christian

 

[tom]

@Tom

But you should think of that!

When you update a HACMP-Cluster in an productiv environment you start with
updating the standby-node, moving all resources to the standby node and then
updating the primary node.
After that you move back the resources based on the policy. (cascading / rotating)

Normal bussiness.

Normaly you wont get the downtime from the customer to upgrade both nodes at once.

Regards

Christian

fyi, our new HA cluster will be a complete new design.

 

[mrshark]

what's version 1.5? i do not see it anywhere on the site, nor on the roadmap...

 

[tom]

what's version 1.5? i do not see it anywhere on the site, nor on the roadmap...

See http://www.proxmox.com/forum/showthread.php?t=2853

1.5 is the current testing/beta release.