Race condition related to refresh_interval

[kelchm]

I seem to be encountering a race condition related to configuration reloading when refresh_interval is set to the catalog default of 10 seconds.

Test Configuration

I'm testing in a Rancher environment with three hosts. Each host is running one instance of rawmind/alpine-traefik:1.1.2-1. I am automating deployments to this environment from GitLab which often results in multiple stacks being created/destroyed in a very short period of time.

docker-compose.yml

version: '2'
services:
  traefik-conf:
    image: rawmind/rancher-traefik:0.3.4-18
    network_mode: none
    volumes:
    - /opt/tools
    tty: true
    labels:
      io.rancher.scheduler.affinity:host_label: traefik_lb=true
      io.rancher.container.start_once: 'true'
      io.rancher.scheduler.affinity:container_label_ne: io.rancher.stack_service.name=$${stack_name}/$${service_name}
      io.rancher.scheduler.global: 'true'
  traefik:
    image: rawmind/alpine-traefik:1.1.2-1
    environment:
      CONF_INTERVAL: '10'
      TRAEFIK_ACME_EMAIL: mkelch@foo.com
      TRAEFIK_ACME_ENABLE: 'false'
      TRAEFIK_ACME_ONDEMAND: 'false'
      TRAEFIK_ACME_ONHOSTRULE: 'false'
      TRAEFIK_HTTPS_ENABLE: 'false'
      TRAEFIK_HTTPS_PORT: '443'
      TRAEFIK_HTTP_PORT: '80'
    tty: true
    volumes_from:
    - traefik-conf
    ports:
    - 8000:8000/tcp
    - 80:80/tcp
    - 443:443/tcp
    labels:
      io.rancher.scheduler.affinity:host_label: traefik_lb=true
      io.rancher.sidekicks: traefik-conf
      io.rancher.container.hostname_override: container_name
      io.rancher.scheduler.global: 'true'

Symptoms

Traefik acts as if no configuration change has occurred. About a third of my deployments and up being unreachable (traefik responds with a 404 error) until I upgrade the Traefik services to force a reload.

What I've tried

I've had success in changing the refresh_interval to be less frequent. With a setting of 60 seconds I have not had a single occurrence of this issue so far.

[kelchm]

After digging into this a bit more it's clear that the rules.toml file is being updated correctly, so any issue must be an upstream issue with traefik itself.

I'm going to go ahead and close this issue out.