Troubleshooting › CPU Usage Spikes in server nodes with Secured Simplified Communication and Load Balancer

CPU Usage Spikes in server nodes with Secured Simplified Communication and Load Balancer

Symptom:

When a clustered Orchestrator is configured to use secure (HTTPS) communication and agents are configured to use simplified communication with an NGINX load balancer, CPU usage spikes can occur in server nodes when the orchestrator is restarted while multiple agents are active.

Solution:

Note: This is not a recommended approach. Use the following approach only for the above problem.

To fix the problem, configure the clustered Orchestrator to use non-secure (HTTP) communication between load balancer and orchestrator as follows:

1. Add the following properties in the OasisConfig.properties file for all orchestrator nodes:
   • pam.transport.unsecured.connector.flag: Set the value to true.
   • pam.transport.unsecured.connector.port: Specify an integer value. Default value is 80.
2. Edit the following entries in the secure-pam-server.conf file to enable non-secure websocket connection between Load Balancer and nodes:
   • // websocket connections

     upstream node1{

     server <hostname of machine where you have installed node1>: Specify the same port number as pam.transport.unsecured.connector.port

     }

     Note: Specify the same port number as pam.transport.unsecured.connector.port for upstream node2.

   • Change https to http in the following entries:

     location = /ws/node1 {

     proxy_pass http://node1; }

     location /ws/node1/ {

     proxy_pass http://node1;}

     location = /ws/node2 {

     proxy_pass http://node2; }

     where node2 is the upstream name

     location /ws/node2/ {

     proxy_pass http://node2; }

3. Restart Orchestrator nodes and Load Balancer.