Group Start and Stop Order:
Notes:
- You can't use 'pcs -f cibname' for this command, it will say "Group not Found"
- This will start groups in order input and stop them in reverse order
pcs constraint order start SQL-Group then WWW-Group pcs constraint order start WWW-Group then ASTERISK-Group
Colocations Run on same Node
Infinity = MUST run on same node
pcs constraint colocation add WWW-Group with ASTERISK-Group INFINITY
Disable STONITH (Shoot the Other Node in the Head)
pcs property set stonith-enabled=false
Prefer node1 for webserver :
pcs constraint location webserver prefers node1=50
Set Boot Order virtual_ip then webserver:
pcs constraint order virtual_ip then webserver
Require webserver to run on same host as virtual_ip
pcs constraint colocation add webserver with virtual_ip INFINITY
Change timeout from 20 seconds to 240
pcs resource op defaults timeout=240s
Based on ClusterLabs reccomendation regarding how to handle data after a recovery, they recommend using the following.
pcs resource defaults resource-stickiness=100 && pcs resource defaults
I rec clicking the link above to further understand what is going on with that command.
Remove Resource :
pcs resource delete ClusterMon-External
Remove Constraint :
1st gather info :
pcs constraint --full
#(id=location-webserver-node2-50)
pcs constraint remove location-webserver-node2-50
Re-Check:
pcs constraint --full
Good Reads :