Why not just do DB replication?
On 18 Oct 2017 10:33 pm, "Sud Ramasamy" <to_sud(a)yahoo.com> wrote:
Hi Keycloak devs/users,
We are trying to wrap our heads around how we might deploy Keycloak in a
PROD with DR failover topology and are running into a concern with the
client secret being different between the clusters.
We have two separate Keycloak clusters with their own databases for our
PROD and DR datacenters. As part of initial one-time client setup when we
register the client in the PROD cluster we also register the same client in
our DR cluster. The configuration for the client is identical between the
two clusters except for the client secret which is generated by Keycloak.
When there is a DR event for Keycloak (either failure or scheduled
maintenance) we have the ability to repoint the URL for the PROD Keycloak
to the DR Keycloak cluster. We don’t change anything else. Unfortunately
the PROD clients will not be able to establish SSO with the DR cluster
because the client secret is different.
We’ve considered instead of using the Keycloak APIs to register the client
in both clusters (thereby having different client secrets) to register the
client in one cluster and use database scripts to push the same
configuration to the other cluster database and thereby keep the secrets
the same.
I was wondering if others have run into this limitation and how you may
have solved for it. Also we are on Keycloak 2.5 (for RH-SSO support
purposes). This might be addressed in the upcoming release of Keycloak with
multi-datacenter support. But that is currently not an option for us.
Thanks in advance for your insight.
-sud
_______________________________________________
keycloak-user mailing list
keycloak-user(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/keycloak-user