Rancher/Server Outage & Reconnection

Context

  • Migrating from Docker Cloud.
  • Have a few questions.

Problem

  • I have my rancher/server node in an ASG.
  • For some reason the node dies and a new one is brought online.
  • The hostname remains the same as an internal ELB is pointing at the ASG.

Questions

  1. Do the agents continue to function during the outage?
  2. Will the agents automatically be able to “talk” to the new server node or will they have to “register” again?

Thanks!

@revett

  1. You’ll need the database stored somewhere externally, like RDS. If the DB is in the dead server starting up a new one with an empty DB will not bring things back the way they were.

0.5) There will be a simpler way to setup HA with multiple servers before GA at the end of the month.

  1. The agents themselves wouldn’t be doing anything other than trying to reconnect, but your containers on the host and networking between them continue to work. If a host/container dies while there is no server up it won’t be rescheduled or removed from DNS until there is a server again.

  2. Yes, as long as the host registration setting is set and the hosts were registered using that hostname.

1 Like