google cloud health check failing

1
votes

I have tried everything I could find on google but still can't make it work. I configured health checks on port 8080, I have created nodejs server to serve 200 OK and close the connection right away (no keep alive). Manual tests to the ip address of the instance seems to work. I used ngrep to check incoming connections on port 8080 and I also print to the console from node and see nothing, no traffic to port 8080. Also on the firewall I have a rule to allow port 8080 on 0.0.0.0/0 I tried on port 80 in the beginning and switched to port 8080 just for the sake of trying to make it work. I found out that "google-address-manager" service was failing and started it, also tried to reboot the instance. I also tried adding 130.211.0.0/32 to the local routes with the command: sudo /sbin/ip route add to local 130.211.0.0/32 dev eth0 proto 66 I tried network load ballancer and http load ballancer with same result. Can you please help me? I've been trying for the last 72 hours and can't make it work.

Here's a copy paste of the health check definition: default-health-check In use by
lb-please-work-backend-service Path
/ Port
8080 Interval
5 seconds Timeout 25 seconds Unhealthy threshold 20 consecutive failures Healthy threshold
1 consecutive success

**update: I've now created a new server from scratch, with nodejs listening on port 80 and returning 200 OK. I also created a new LB and healthchecks and everything and still nothing

1
node.jscloudload-balancinggoogle-compute-enginegoogle-cloud-platform
Can you provide more information? It's like a black box at the moment for us. - EWit
Can you try changing the default index page on your web server? just add anything to it and see if it still gonna fail - George
Can you SSH into you instance and run the tcpdump command as follows: tcpdump -A -n host 169.254.169.254 and provide the output? We're loooking to see how the health check requests that come from the LB are being handled. - Marilu

1 Answers

0
votes

If accessing the page on your instance works just fine but no the Load Balancer (LB), is very likely that the requests coming from the LB aren't being closed properly, therefore the instance is unhealthy. I'll recommend capturing a tcpdump file to find out more about it.