The proper procedure to shut down a Clustrix Appliance node is as follows:
- shutdown node with "shutdown -h now".
- remove cables from node and remove node from rack
- move rails
- re-plug node in to infiniband and ethernet
- re-plug node into power
- power node on
- wait for node to re-join cluster and full protection restored for all data alert to be sent
Note that shutting down the node and powering the node back on will both result in a group change.
By default the cluster will wait for 10 minutes before reprotecting around a removed node. If you are planning on re-incorporating the node you should endeavor to have the node back online before this time expires. If you need more than 10 minutes you can change this with a global variable.
The global variable that defines how long the cluster waits for a node to come back before it starts the reprotection process is:
rebalancer_reprotect_queue_interval. This is by default set to 600 seconds (10 minutes).
If you want to change this, you can do this with:
set global rebalancer_reprotect_queue_interval=1200;
or some other value in seconds.