0

I have topology looks like

User -> DNS -> Haproxy (google compute) -> Google Load Balancer -> Instance Group

The traffics is comming to haproxy and mapping to 3 load balancers.

I get a lot of responses with -1 code on haproxy, for example

0/0/1/-1/287 -1
0/0/6/-1/109 -1
0/0/2/-1/233 -1

I read from haproxy docs 8.5 section, the docs said :

enter image description here

I want to know why so many get timeout server, because instance group under load balancer is fine. How can I trace them because all request with that response not comming to each instance on instance group?

itx
  • 99
  • 1
  • This is just a suggestion, but in AWS, i usually get time out when ports are closed on the instance side or loadbalancer side, try rechecking the ports, both inside the instance and on the console. – logax Jul 05 '20 at 20:44

1 Answers1

1

Could you confirm you're using a HTTP(S) load balancer? If so, check Stackdriver Logging and see if there's any 502 error codes and what are the response codes?

Also have you tried increasing the timeout server yet? https://cbonte.github.io/haproxy-dconv/1.7/configuration.html#timeout%20server

EliasH
  • 21
  • 2
  • Yes I'm using http(s) LB, there is no any 502 error code on the logs, I have set `timeout connect` 5 seconds, `timeout client` 50 seconds, `timeout server` 50 seconds – itx Jul 08 '20 at 09:01