0

I'm using GCP Loadbalancer for accessing my web applications. There are some cases where my service might take sometime to give the response back (more than 300 seconds). When I check with GCP settings I saw two timeouts (Connection Timeout and Connection Draining Timeout) as mentioned in the below link,

GCP load balancer 502 server error and "backend_connection_closed_before_data_sent_to_client" IIS 10

Even after increasing the time out, still my requests are getting timedout at 300 seconds. Is there any other time out that has to be configured? Or do I need to configure TCP for the HTTPS GCP Loadbalancer which I have now.

Operating System : Windows Server 2012 Web Server: WildFly

Please give some suggestions. Thanks in Advance.

Anton Reema
  • 3
  • 6

1 Answers1

1

HTTP 502 means that the thing your LB was connected to -- your backend service -- disconnected suddenly, so the problem is not between the client and LB, but between LB and your service. You'd see this picture if your backendservice crashes. Since it's happening exactly 300 after you start your request it is likely that there's some other timeout in play, but it's somewhere in LB <-> backend service.

[EDIT] I had a similar issue when our LB was timing out after 300 seconds. When I changed the timeout setting in the Web UI it was getting reset back to 300 seconds after a minutes or so. It turned out it was set by our Kubernetes -- when one configures BackendConfig a special controller is installed that keeps track of the timeout setting and one needs to change the timeout value appying new BackendConfig config, i.e., not in Web UI.

Anton Daneyko
  • 6,528
  • 5
  • 31
  • 59