Dumb questions regarding HA

[AlexandreAngel]

Hello,

I did setup a cluster with two nodes (I know it's bad and i understand why) using two_node: 1 (without it, my dear server were rebooting when his brother was missing, after a clean reboot).

I tried setting up a ha group with my two nodes and adding a vm as ressources.
The vm is started where it should. But when i reboot the node, the vm stay down until the node gets back without starting it on the other node.

Is it the normal behavior ?

When a node is missing (a clean shutdown), i can't start the vm hosted on that node cause i need to migrate them and i get a no route to host.
The vm config file is in /etc/pve/nodes/nodename/qemu-server/id.cfg and is accessible by the online node.

What do i have to do to start this VM ?

 

[dietmar]

Never use HA with 2 nodes! That does not work.

For HA, you need at least 3 nodes.

 

[AlexandreAngel]

Can i have 3 servers but only two are members of an ha group that run the vms ?
Can i have a cluster of two nodes without HA ?
How i disable HA ? It seams the system adds HA when you create a cluster.

 

[dietmar]

Can i have 3 servers but only two are members of an ha group that run the vms

yes

Can i have a cluster of two nodes without HA ?

yes

How i disable HA ? It seams the system adds HA when you create a cluster.

No need to disable. Simply do not enable/use HA for VMs.

 

[AlexandreAngel]

With no VMs with HA enable and without the two_node: 1. The first node of my cluster is rebooting when i reboot the second node.

And from my first post :

When a node is missing (a clean shutdown), i can't start the vm hosted on that node cause i need to migrate them and i get a no route to host.
The vm config file is in /etc/pve/nodes/nodename/qemu-server/id.cfg and is accessible by the online node.

What do i have to do to start this VM ?

 

[dietmar]

With no VMs with HA enable and without the two_node: 1. The first node of my cluster is rebooting when i reboot the second node.

not sure what you mean by "no VMs with HA enable" - the watchdog is not enabled in that case, so it will not reboot.

When a node is missing (a clean shutdown), i can't start the vm hosted on that node cause i need to migrate them and i get a no route to host.
The vm config file is in /etc/pve/nodes/nodename/qemu-server/id.cfg and is accessible by the online node.

What do i have to do to start this VM ?

You need to gain quorum, for example using:

# pvecm expected 1

But first make sure the other node is offline.

 

[AlexandreAngel]

I meant that there is no Ressources listed in the HA part of the "Datacenter".

Concerning the other part, how to i get this vm with ID 113 to run on another node using the web UI ? I can't migrate it.

I want to have 2 nodes with the ability to migrate VMs between them :
- I create a cluster on node 1
- I add node 2 to node 1
- Reboot everybody
- I do not enable HA by creating a HA group and adding VMs to HA ressources.

Is there something else to do or not to do ?

 

[dietmar]

see https://pve.proxmox.com/pve-docs/pve-admin-guide.html#_recovering_moving_guests_from_failed_nodes