Arseniy Tashoyan created FLINK-29206:
----------------------------------------
Summary: Default Flink configuration contains irrelevant settings
Key: FLINK-29206
URL: https://issues.apache.org/jira/browse/FLINK-29206
Project: Flink
Issue Type: Bug
Components: Kubernetes Operator
Affects Versions: 1.15.2
Reporter: Arseniy Tashoyan
The default configuration of a FlinkDeployment contains the following settings:
{code:yaml}
kubernetes.operator.metrics.reporter.slf4j.interval: 5 MINUTE
kubernetes.operator.metrics.reporter.slf4j.factory.class:
org.apache.flink.metrics.slf4j.Slf4jReporterFactory
kubernetes.operator.reconcile.interval: 15 s
kubernetes.operator.observer.progress-check.interval: 5 s
{code}
These settings are specific to Flink Kubernetes Operator itself, but not to
Flink applications that the Operator runs. Although these settings most
probably do not make any influence on a Flink application (they are just
ignored), but the configuration is misleading. It would be better to remove
these settings from the default configuration of a Flink application.
Actually these settings come from the _defaultConfiguration_ section of the
file {_}helm/flink-kubernetes-operator/values.yaml{_}:
{code:yaml}
defaultConfiguration: ...
flink-conf.yaml: |+
# TODO Remove these Operator-specific settings - they are irrelevant to
Flink apps
kubernetes.operator.metrics.reporter.slf4j.factory.class:
org.apache.flink.metrics.slf4j.Slf4jReporterFactory
kubernetes.operator.metrics.reporter.slf4j.interval: 5 MINUTE
kubernetes.operator.reconcile.interval: 15 s
kubernetes.operator.observer.progress-check.interval: 5 s
{code}
This piece of configuration is misleading - is it the configuration of the
Operator itself or the configuration of applications started by the Operator?
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
