Closed zdykstra closed 2 years ago
Hi @zdykstra , thanks for opening the issue.
I've tried to reproduce the issue on my machine following the steps described on Running a Keycloak Cluster using JDBC_PING in a local Docker network
In my experiment, I've started and shut down, several times, two keycloak-clustered
instances and, at the same time, checking the records in JGROUPSPING
table.
Indeed, I can see the error logs you mentioned above but, checking the records in JGROUPSPING
, it kept correctly according to the number of keycloak-clustered
instances at a specific time. When one was shut down, its record was deleted from JGROUPSPING
table although the error message was logged anyway.
So, unfortunately, I couldn't reproduce your case.
Let me know if I can help in another way, maybe trying a different approach.
Best regards,
@ivangfr thank you for taking the time to look at this and attempting to replicate the problem, I appreciate it. I've also attempted to replicate this again on GCP with the keycloak-clustered 15.0.2 container and a hosted pgsql database instance. Instances are currently being removed from the jgroupsping
table on shutdown. So right now, the problem is either resolved with 15.0.2 or isn't readily repeatable.
I noticed that I have a number of entries in the
jgroupsping
table under Postgres, even after the container has been shutdown and destroyed. Looking at the docker logs for a shutdown container, I see this:Based on this, it looks like
[java:jboss/datasources/KeycloakDS]
is being shutdown before the finalorg.jgroups.protocols.JDBC_PING
deletion takes place.