I am setting up Keycloak v19 in a 3 node cluster with my own custom docker image with a shared database. There is a problem with the distributed cache as the nodes cannot communicate with each other. This is in Azure running in App Service Environment (v3). The issue seems to be related to the ASE nodes having ports locked down. I want to completely disable distributed cache and always perform database lookups whilst the issue is investigated and use sticky sessions using AUTH_SESSION_ID cookie configured at the load balancer.
Apart from the obvious performance hit, are there any other downsides to doing this, for example actionTokens. I don’t see any as the requests (e.g. for reset password requests) should hit the same back-end and request replay protection should still be acheivable.
What is the best way to completely disable the cache, or should I continue to use local cache but disable realm and user cache items only ?