Node maintenance mode UI

[SteveITS]

Is your VM subject to HA? Logs? It should migrate if it’s not stopped.

 

[pulipulichen]

I've tried the maintenance command :
ha-manager crm-command node-maintenance enable pvbe20

CT migrated well but the VM is not moving. Normal behavior ? Why ?

Hi Singman,

This is indeed the expected behavior. The ha-manager command and the maintenance mode it triggers only apply to resources that are currently managed by the High Availability (HA) stack.
If your VM is not added to the HA resources list, the CRM (Cluster Resource Manager) will not track its state or automatically migrate it when the node enters maintenance. For any VMs or CTs that are not under HA management, you will still need to migrate them to another node manually before the maintenance can proceed fully.

 

[Elompenta]

I've tried the maintenance command :
ha-manager crm-command node-maintenance enable pvbe20

CT migrated well but the VM is not moving. Normal behavior ? Why ?

This depends on your HA config. But please create a new thread. The topic here is to show this feature in the UI or not

 

[h@rald]

I feel that the strategy to only migrate VM's added to HA when a node is put in maintenance might need a review not that dynamic load balancing is a thing.
DRS should be able to recognize a host in maintenance mode and migrate all running machines to an available node.
(whilst complying with DRS restrictions and pinning of course)

And please implement a GUI way of assigning maintenance mode to a host.

 

[h@rald]

I feel that the strategy to only migrate VM's added to HA when a node is put in maintenance might need a review not that dynamic load balancing is a thing.
DRS should be able to recognize a host in maintenance mode and migrate all running machines to an available node.
(whilst complying with DRS restrictions and pinning of course)

And please implement a GUI way of assigning maintenance mode to a host.

 

[Madozu]

+1 for an easy way to enable a node's maintenance mode. On the node's top line in the right pane (besides the "Shutdown" and "Reboot" buttons) would be a nice place. Or in the node's context menu (right click).

And as previously stated: this should also pause the replications to/from that node.

Why would that be a very helpful additon?

• Easy "push button" to evacuate a host
• Maintaining the hardware / UEFI / Adapter firmware often requires several reboots without wanting the workload back immediately
• Manually finding / disabling the replication jobs is error-prone
• Isolate host-specific problems (by making sure that all workload is evacuated)
• ... because VMware has it too ...