- proxpatch is used to upgrade your proxmox nodes with apt-get upgrade while taking care the cluster(proxmox + ceph) stays available.
- in general folllowing sequence is applied:
- check prerequisites - cluster health, HA group assignment
- set ceph maintenance mode - patch cluster node by node and take care that the running vms are migrated to an other node
- use a separate management machine to spin the script (dont use a proxmox clusternode!)
- setup ssh trust between clusternodes an management machine.
- fill in your node Information in proxbatch.cfg
- CLUSTER=[clustername]
- NODES=[amount of nodes]
- NODE1_IP=[IP Address]
- NODE1_USER=[Proxmox admin user]
- NODE1_NAME=[Proxmox node name]
- NODE1_HAPRIOGRP=[name HA Prio Group]
- NODE2_IP=[IP Address]
- NODE2_USER=[Proxmox admin user] ....
- NODE[n]_IP=[amount of nodes]
- NODE[n]_USER=[Proxmox admin user]
- NODE[n]_NAME=[Proxmox node name]
- NODE[n]_HAPRIOGRP=[name HA Prio Group]
- every running machine has to be assigned to an HA Group
- there must be a ha group for each Cluster node which priorize that Node
- priorize node1 in HA Group for node1 like this : node1:3,node2:2,node3:2
- ceph Pool size has to be setup to tolerate a node shutdown (eg. size 2/1)