PVE2ClusterTestBed: Difference between revisions

From Proxmox VE
Jump to navigation Jump to search
Line 53: Line 53:
==== node A :====
==== node A :====
  pvecm add <ip of node B>
  pvecm add <ip of node B>
if you have error "unable to copy ssh ID", do next step and retry, else skip next step
'''if you have error "unable to copy ssh ID", do next step and retry, else skip next step'''


==== if "unable to copy ssh ID" - node B :====
==== if "unable to copy ssh ID" - node B :====

Revision as of 08:02, 9 July 2013

initial setup

  • two nodes (A and B), with one admin network and one DRBD network configured
  • drbd configured
  • one KVM node id 100 installed on a LVM over DRBD, with no virtual CDROM inserted
  • node 100 live migration OK between the two nodes
  • backup of /etc/pve in /root/backup/pve

delete cluster configuration, on both nodes :

service pvestatd stop
service pvedaemon stop
service pve-cluster stop
umount /etc/pve
/etc/init.d/cman stop
rm /etc/cluster/cluster.conf
rm -rf /var/lib/pve-cluster/*

wait 20 seconds if the server has just restarted

service pve-cluster start
service cman start
service pvestatd start
service pvedaemon start

put the nodes in a cluster

node A :

pvecm create <cluster name>

if backup of previous /etc/pve configuration exists :

 cd /root/backup/pve
 cp storage.cfg /etc/pve/
 cp nodes/<source node name>/qemu-server/100.conf /etc/pve/nodes/<node A name>/qemu-server/

else, configure DRBD storage, and install VM 100 Log into PVE WEB interface, verify storage, start VM 100, vm100 started.

node B :

pvecm add <ip of node A>

Log into PVE WEB interface, verify storage, live migrate VM 100, live migrate back to node A.

node A power failure, VM restart on node B, service OK

Remove power plug from node A (or push >4s power button)

node B :

pvecm nodes #shows node A as status "X"
pvecm expected 1
mv /etc/pve/nodes/<node A name>/qemu-server/*.conf /etc/pve/nodes/<node B name>/qemu-server/

Log into PVE WEB interface, start VM 100.

node A reinstallation

Install node A as in initial setup, without cluster configuration : to do this quickly in test platform :

  • stop node B (halt)
  • start node A
  • follow "delete cluster configuration, on both nodes", but just on node A
  • regenerate ssh keys on node A :
/bin/rm /etc/ssh/ssh_host_* && dpkg-reconfigure openssh-server
  • start node B

node A :

pvecm add <ip of node B>

if you have error "unable to copy ssh ID", do next step and retry, else skip next step

if "unable to copy ssh ID" - node B :

pvecm expected 1

verification

on node A and node B, verify that the status is "M" for both nodes :

pvecm nodes

verify you can log into PVE WEB interface of the two nodes, verify you can live migrate