After-hours switchover

checklist.md

Rolling forward

• Set k8s load balancers to listen on ports 81 and 444:
  • welcome (a6478641c8f4411e599d106fa7745be6-898755358.us-west-1.elb.amazonaws.com)
  • huyang (a6442f4148f4411e599d106fa7745be6-1097607471.us-west-1.elb.amazonaws.com)
  • tenants (a6456bf648f4411e599d106fa7745be6-289499823.us-west-1.elb.amazonaws.com)
  • messaging (a518bc87b8f4411e599d106fa7745be6-242074411.us-west-1.elb.amazonaws.com)
  • huyangapi (ac85c06dd8ca111e599d106fa7745be6-165125648.us-west-1.elb.amazonaws.com)
• pg_dump gridium from a k8s minion
• docker-compose stop on docker1 and docker2
• psql -h $EXISTINGRDS < dump.sql
• adjust Route53 entries to point to new ELBs
  • welcome.gridium.com -> docker-welcome
  • *.tikkit.gridium.com -> docker-huyang
  • *.tikkit.us -> docker-tenants
  • messaging.gridium.com -> docker-messaging
• Test after propagation
  • welcome.gridium.com
  • internal.tikkit.gridium.com
  • internal.tikkit.us
  • messaging.gridium.com (http & https)

Rolling back

In case something bad happens.

• Re-set route53 to point to original ELBs
• Change ports back to 80 and 443
  • welcome (a6478641c8f4411e599d106fa7745be6-898755358.us-west-1.elb.amazonaws.com)
  • huyang (a6442f4148f4411e599d106fa7745be6-1097607471.us-west-1.elb.amazonaws.com)
  • tenants (a6456bf648f4411e599d106fa7745be6-289499823.us-west-1.elb.amazonaws.com)
  • messaging (a518bc87b8f4411e599d106fa7745be6-242074411.us-west-1.elb.amazonaws.com)
  • huyangapi (ac85c06dd8ca111e599d106fa7745be6-165125648.us-west-1.elb.amazonaws.com)