Your upgrade to Kublr 1.20.2 must include the procedures described below.
When upgrading a Kublr Control Plane, first upgrade the following fieature versions in the KCP cluster specification simulatzniously and wait for the KCP to update and recover:
After that you can update log collection and monitoring feature versions via UI.
Before updating log collection feature, go to Kibana Management menu and delete kublr*, kublr_default* index patterns
After the new cluster specification update is applied and the helm charts are installed or upgraded, run the following command from the logging-controller pod:
/sgadmin/tools/sgadmin.sh --delete-config-index -icl -nhnv \
-cacert /sg-certificates/root-ca.pem \
-cert /sg-certificates/sgadmin.pem \
-key /sg-certificates/sgadmin.key \
-h $ELASTICSEARCH_DISCOVERY_SERVICE
Trigger kublr-logging-sg-init
CronJob in kubernetes dashboard or use the following commnand:
kubectl create job kublr-logging-sg-init --from=cronjob/kublr-logging-sg-init -n kublr
Starting from Kublr 1.20.2, Hot/Warm/Cold nodes are supported for Elasticsearch, which led to updated structure of the logging feature values.
If custom values are used for logging feature in the Kublr cluster specification, it needs to be modified to follow the new structure:
data/master nodes’ configuration parameters are moved to the nodeGroups
section
xpackEnabled
flag is used to enable Elasticsearch XPack
Here is an example of the new cluster specification values structure for logging:
logging:
values:
elasticsearch:
xpackEnabled: true
nodeGroups:
data:
persistence:
enabled: true
storageClass: thin-disk
master:
persistence:
enabled: true
storageClass: thin-disk