Hello
I am after some advice
We are looking at when nodes fail in AWS, our plan is that is a node fails to spin up a new node with a custom Edge AMI. This would be at the stage just before the install component process is done
What i am trying to work out is that best way to deregister the failed node (lets say it has a Router and a Message Processor)
Do any of you do this already?
SO an example
We have 3 nodes each with a RMP install
One node dies
New node spun up from AMI and Script kicked off to install RMP
Once installed the normal process is followed and the install registers the RMP with the MS
Now we have a dead node that still has a registered RMP with the management server
Whats the best way to have this automatically removed?
The same situation would happen if we used autoscale, once the load went back down we would want to remove the autoscaled node and deregister it form the MS
If there is a better way to do this i am all ears 🙂
Thanks
@Michael McLean Are you looking for the APIs to deregister the Router and MP or the sequence of steps?
If you are looking for APIs this http://docs.apigee.com/private-cloud/latest/removing-server-management-servermessage-processorroute... has it.
Sequence:
Take the old RMP out of rotation using this http://docs.apigee.com/private-cloud/latest/enablingdisabling-server-message-processorrouter-reachab....
Add the new RMP with the steps mentioned here http://docs.apigee.com/private-cloud/latest/adding-router-or-message-processor-node
Deregister the old RMP, ref: http://docs.apigee.com/private-cloud/latest/removing-server-management-servermessage-processorrouter
Validation: (Run before and after to see the change)
Curl -v http://mgmthost:8080/v1/servers?pod=gateway®ion=dc01
Curl -v http://mgmthost:8080/v1/servers?pod=analytics®ion=dc01
@Michael McLean in addition to the steps described by @Maruti Chand you need to remember to remove the MPs from the existing environment(s) and add new MPs to the environment(s)
User | Count |
---|---|
7 | |
2 | |
2 | |
1 | |
1 |