How to persist user session at keycloak running on kubernetes? - kubernetes

Question

After pod restart, all user session data are missing.
But all other data are present after pod restart (ex:- realm, user, realm settings)
keycloak is running with Postgres as persistence storage in a single pod.
below is the deployment file config:
apiVersion: apps/v1
kind: Deployment
metadata:
name: idms
namespace: default
labels:
app: idms
spec:
replicas: 1
selector:
matchLabels:
app: idms
template:
metadata:
labels:
app: idms
spec:
containers:
- name: postgres
image: registry.prod.srv.da.nsn-rdnet.net/edge/postgres:12.3-alpine
imagePullPolicy: "IfNotPresent"
ports:
- containerPort: 5432
lifecycle:
postStart:
exec:
command: ["/bin/bash","-c","sleep 5 && PGPASSWORD=$POSTGRES_PASSWORD psql $POSTGRES_DB -U $POSTGRES_USER -c \'CREATE SCHEMA IF NOT EXISTS keycloak;\'"]
envFrom:
- configMapRef:
name: postgres-config
- name: keycloak
image: quay.io/keycloak/keycloak:10.0.1
env:
- name: KEYCLOAK_USER
value: "XXXXXXX"
- name: KEYCLOAK_PASSWORD
value: "XXXXXXX"
- name: REALM
value: "XXXXXXX"
- name: PROXY_ADDRESS_FORWARDING
value: "true"
- name: DB_VENDOR
value: "POSTGRES"
- name: DB_ADDR
value: "localhost"
- name: DB_PORT
value: "5432"
- name: DB_DATABASE
value: "postgresdb"
- name: DB_USER
value: "xxxxxxxxx"
- name: DB_PASSWORD
value: "xxxxxxxxx"
- name: DB_SCHEMA
value: "keycloak"
- name: KEYCLOAK_IMPORT
value: "/opt/jboss/keycloak/startup/elements/realm.json"
volumeMounts:
- mountPath: /var/lib/postgresql/data
name: postgredb
- mountPath: /opt/jboss/keycloak/startup/elements
name: elements
ports:
- name: http
containerPort: 8080
- name: https
containerPort: 8443
readinessProbe:
httpGet:
path: /auth/realms/master
port: 8080
volumes:
- name: elements
configMap:
name: keycloak-elements
- name: postgredb
persistentVolumeClaim:
claimName: postgres-pv-claim
can you please let me know, any configuration required to persist user session?

Answer 1

Please take a look at the existing chart.
On High Availability and Clustering section, it shows you that you need to set: CACHE_OWNERS_COUNT to 2 or higher and CACHE_OWNERS_AUTH_SESSIONS_COUNT to 2 or higher.
It's also recommended to use a separate Infinispan cluster if you have a big (more than 1 million from my experience) sessions.