Behaviour of the Liferay Infrastructure on LXC-SM when a Headless API is called during deployment

Issue

1. What will be the behavior of the Liferay Infrastructure on LXC-SM when a Headless API is called via 'Postman' or any other third-party system while a deployment is going on?

2. Will the requests always be served, or will they fail or be diverted to another pod when one is deployed?

Environment

  • Liferay DXP 7.1
  • Liferay DXP 7.2
  • Liferay DXP 7.3
  • Liferay DXP 7.4

Resolution

    • During the deployment, it’s expected to see minimal downtime as the traffic should be routed to the healthy backend services. Also, by default, the services in Liferay Cloud are pre-configured to use the RollingUpdate strategy, it will allow deployments to take place with zero downtime by incrementally updating instances with new ones.
    • For the requests that pass through the external load balancer have the following route ELB(Ingress) -> Nginx -> Haproxy -> Liferay, the client session will be maintained as a sticky session at HAProxy and then passed to the healthy backend service if one of the backend services crashes/stops due to some reason.
    • However, the request goes through ILB(that comes through VPN or third-party system) or Postman doesn’t have any extra layer. During the deployment, ILB forwards the requests to the healthy backend service, the requests will be served else they will fail due to the pod scheduling and initialization as the ILB continuously performs health checks on all backends and will route them accordingly to minimize the disruption.
    • When the connection comes to Liferay through the Site-to-Site VPN, the following route is established:
      Customer premises -> GCP Firewall -> ILB (Internal Load Balancer) -> Endpoints (liferay, database, webserver)

Additional Information

Was this article helpful?
0 out of 0 found this helpful