HAProxy crash/restart issue Rancher v0.51.0

Hi folks,

I managed to catch an issue and need some help. My setup is an Amazon ELB -> Docker/Rancher hosts with load-balancers -> docker containers.

For whatever reason, HAProxy crashed on 1 host, which happened to be the next host for the ELB. That resulted in a 503 Service Unavailable: Back-end server is at capacity.

Two questions:

  1. How do I troubleshoot this, since Rancher deletes the loadbalancer container on failure?
  2. Is this an existing problem that would be fixed by upgrading to v0.56.1? In looking at the release notes, it seems this might be improved, but I’m not positive.

It sounds as if you could be hitting this issue, which will be in the next release.