Question
Accenture
US
Last activity: 29 Apr 2019 10:41 EDT
Mark Node Offline for Load Balancer Prior to Quiesce
Hello,
I have noticed that the ping monitor rest service (accessible from: https://hostname.example.com/prweb/PRRestService/monitor/pingService/ping ) has been updated in Pega 8.2 to include more information and more importantly a "state" exported in JSON format.
Hello,
I have noticed that the ping monitor rest service (accessible from: https://hostname.example.com/prweb/PRRestService/monitor/pingService/ping ) has been updated in Pega 8.2 to include more information and more importantly a "state" exported in JSON format.
I am trying to figure out where this is configured in Pega. I only see the pyGetHealthStatus activity which was used in Pega 7.
Can you help to figure out where this JSON is configured? Also, is there a way to administratively manipulate this information? I want to mark a node as unavailable so a load balancer does not send any additional traffic to that node prior to performing a quiesce via an API call.
Thanks,
Jeff
***Edited by Moderator Marissa to update platform capability tags; add SR Details****
Pegasystems Inc.
US
Hi,
Not sure you want to use Pega ping service to mark down a node, that is the task of LB. For example,
https://support.f5.com/csp/article/K13310
In Websphere IHS plugin (which is a lightweighted LB), you can simply set the loadbalancerweight to zero for a specific Websphere node.
Accenture
US
Thanks. We're using an AWS Classic ELB. However, we don't have full control of our AWS Infrastructure. There is another group in our organization that handles this. I was looking for a way that we could manage it and work with the current Load Balancer rules. We can have them configure the Load Balancer to Deregister an Instance and/or Configure Slow Draining, but since it is managed by another group, we need to put in a ticket to have this completed. We can go down this route if it is the best suggestion.
Do you have any other suggestions on handling this either from an API call, or some other authenticated method in Pega?
Pegasystems Inc.
US
You should then have your AWS LB group to use pingservice as healthcheck, similar to what Pegacloud env does. Everything should then be automatic without manual intervention.
Accenture
US
When quiescing a node, do you need to de-register the node from the ELB first? When the cookie is invalidated at the LB, is the health of the pingService for that node changed accordingly or could the load balancer send the user back to the same node?