Rancher UI Behind AWS ALB Issues in 1.2

Before upgrading to Rancher v1.2 we had Rancher v1.1.4 behind a AWS ELB with websockets and proxy-protocol setup. This worked fine and the UI was normal, however after upgrading to Rancher v1.2.2 the UI seems much more temperamental and unresponsive now. I don’t know if it’s the ALB or the 1.2 upgrade however.

  1. Switching Environments in the upper corner doesn’t work, and requires a full page refresh
  2. Stacks don’t refresh fully, and containers in the UI say “starting…” but looking at them from the docker cli shows them up and running
  3. Deployments seem much “slower”, with some stacks that used to come up in seconds in v1.1.4 taking minutes.
  4. Viewing JSON logs in the UI on a running container constantly says “Disconnected”. Occasionally the logs do show up, but it’s inconsistent. Getting a shell on a container works properly.

Infra: Amazon EC2 m4.xlarge instances with 3200iop SSD EBS-optimzed storage
Rancher: v1.2.2
OS: Amazon Linux 2016.09 for agents and master
Docker: v1.11.2 for agents and master
Database: MySQL 5.7 in AWS RDS
Orchestration: Cattle

1 Like