Load balancing - unhealthy instances in google compute engine Network load balancer

Question

I have installed my web application on 2 Windows based VMs of GCE.My application runs on 8080 port.

Steps followed for Netwrok Load Balancer :

1) I created health checks for 8080 port. 2) Added both my VMs and helathchecks to target pool. 3) In forwarding rule I created a rule for 8080 port for that particular Target Pool.

After this go to Target Pools and check the health of the VMs

Here a red symbol is shown against both the instances and message shown as "instance is unhealthy for ".

I have added port 8080 in Firewall rules.

If any one can help, if I am doing anything wrong or there is some other way to setup the Load Balancer.

Answer 1

I believe this issue is not related to the fact that you are listening in port 8080. Health check will pass as long as your instances are able to communicate with the Metaserver (169.254.169.254 [1]) and response with a valid HTTP page.

You must be sure you have allowed communication on port 8080 on the Google Firewall and on your Windows firewall instance [2]. As a debugging you can try to ping the Metaserver and capturing IP packages to confirm if there is a 3 way handshake between the Metaserver and your GCE instance. Additionally you might want to try to do the setup with the same instances on port 80 to confirm if it is actually related to the port.

[1] https://cloud.google.com/compute/docs/metadata [2] https://cloud.google.com/compute/docs/networking