How to replace nodes without reducing redundancy?

piraeusdatastore / piraeus-operator

The Piraeus Operator manages LINSTOR clusters in Kubernetes.

https://piraeus.io/

Apache License 2.0

409 stars 64 forks source link

How to replace nodes without reducing redundancy? #737

Open yeshl opened 5 days ago

yeshl commented 5 days ago

pv with placementCount: "2",how to delete node30 and add a new one to replace it without reducing redundancy?

| master20    | pvc-e8bc738f-633c-4942-83e1-84536c2b7622 | DfltDisklessStorPool |     0 |    1015 | /dev/drbd1015 |            | Unused | TieBreaker |
| node30 | pvc-e8bc738f-633c-4942-83e1-84536c2b7622 | pool-01              |     0 |    1015 | /dev/drbd1015 |  23.34 GiB | Unused |   UpToDate |
| node31 | pvc-e8bc738f-633c-4942-83e1-84536c2b7622 | pool-01              |     0 |    1015 | /dev/drbd1015 |  23.34 GiB | InUse  |   UpToDate |

WanzenBug commented 1 day ago

Simplest would be:

Add the new node
Run kubectl cordon <old-node>
Run linstor node evacuate <old-node>, wait for all resource to sync up.
Delete the node from kubernetes