When each API Connect component looses quorums it changes how the components work.

This is an updated to my 2019 article, which could have been clearer.

API Connect on VMWare

This section covers subsystems that are running on VMWARE.

Gateway

API Management, Portal and Analytics

  • Outage occurs - This is due to API Connect subsystems in VMWare depending on Kubernetes inside of the VMs.

API Connect in OCP/Kubernetes

This article is not going to describe the situation when the the Kuberentes/OCP loses quorum

Gateway (Same as above)

  • without TMS No downtime for your APIs - Gateway will continue to serve traffic
  • without TMS OAuthTokens can be validated
  • with TMS OAuthTokens cannot be validated, and so there may be an interruption for traffic.
  • Unable to publish APIs
  • Unable to store or share revoked OAuth Refresh tokens and Rate Limiting
  • ~Can be fixed by fixed by manually making a remaining node primary~

Manager

  • API Manager will continue to be accessible and usable
  • API Manager will be unable to communicate with the other subsystems

Analytics

  • Access to data stored in the shards on the available nodes
  • Ingestion continues as normal though capacity will be reduced.

Portal

  • Portal is not accessible
  • Unable to Register external consumers
  • Unable to Register new Applications
  • Unable to modify existing assets