docs(autoRestart): document when the restart counter resets for connectors

[rodrigo-molina]

Type of change

• Documentation

Description

While walking through the Kafka Connect AutoReset implementation, I discovered the reset counter feature is not explained in the documentation.

The feature is implemented in:

strimzi-kafka-operator/cluster-operator/src/main/java/io/strimzi/operator/cluster/operator/assembly/AbstractConnectOperator.java

Lines 733 to 754 in 9a0f0ca

	/**
	* Checks whether the connector is stable for long enough after the previous restart to reset the auto-restart
	* counters. Normally, this follows the same backoff intervals as the restarts. For example, after 4 restarts, the
	* connector needs to be running for 20 minutes to be considered stable.
	*
	* @param autoRestartStatus Status field with auto-restart status
	*
	* @return True if the previous auto-restart status of the connector should be reset to 0. False otherwise.
	*/
	/* test */ static boolean shouldResetAutoRestartStatus(AutoRestartStatus autoRestartStatus) {
	if (autoRestartStatus != null
	&& autoRestartStatus.getLastRestartTimestamp() != null
	&& autoRestartStatus.getCount() > 0) {
	// There are previous auto-restarts => we check if it is time to reset the status
	long minutesSinceLastRestart = StatusUtils.minutesDifferenceUntilNow(StatusUtils.isoUtcDatetime(autoRestartStatus.getLastRestartTimestamp()));
	return minutesSinceLastRestart > nextAutoRestartBackOffIntervalInMinutes(autoRestartStatus.getCount());
	} else {
	// There are no previous restarts => nothing to reset
	return false;
	}
	}

and it's used by the handler:

strimzi-kafka-operator/cluster-operator/src/main/java/io/strimzi/operator/cluster/operator/assembly/AbstractConnectOperator.java

Lines 643 to 650 in 9a0f0ca

	// Connector and tasks are not failed
	if (previousAutoRestartStatus != null) {
	if (shouldResetAutoRestartStatus(previousAutoRestartStatus)) {
	// The connector is not failing now for some time => time to reset the auto-restart status
	LOGGER.infoCr(reconciliation, "Resetting the auto-restart status of connector {} ", connectorName);
	status.autoRestart = null;
	return Future.succeededFuture(status);
	} else {

See the full method code:

strimzi-kafka-operator/cluster-operator/src/main/java/io/strimzi/operator/cluster/operator/assembly/AbstractConnectOperator.java

Lines 623 to 664 in 9a0f0ca

	@SuppressWarnings({ "rawtypes" })
	/* test */ Future<ConnectorStatusAndConditions> autoRestartFailedConnectorAndTasks(Reconciliation reconciliation, String host, KafkaConnectApi apiClient, String connectorName, KafkaConnectorSpec connectorSpec, ConnectorStatusAndConditions status, CustomResource resource) {
	JsonObject statusResultJson = new JsonObject(status.statusResult);
	if (connectorSpec.getAutoRestart() != null && connectorSpec.getAutoRestart().isEnabled()) {
	return previousAutoRestartStatus(reconciliation, connectorName, resource)
	.compose(previousAutoRestartStatus -> {
	boolean needsRestart = connectorHasFailed(statusResultJson) || !failedTaskIds(statusResultJson).isEmpty();
	if (needsRestart) {
	// Connector or task failed, and we should check it for auto-restart
	if (shouldAutoRestart(previousAutoRestartStatus, connectorSpec.getAutoRestart().getMaxRestarts())) {
	// There are failures, and it is a time to restart the connector now
	metrics().connectorsAutoRestartsCounter(reconciliation.namespace()).increment();
	return autoRestartConnector(reconciliation, host, apiClient, connectorName, status, previousAutoRestartStatus);
	} else {
	// There are failures, but the next restart should happen only later => keep the original status
	status.autoRestart = new AutoRestartStatusBuilder(previousAutoRestartStatus).build();
	return Future.succeededFuture(status);
	}
	} else {
	// Connector and tasks are not failed
	if (previousAutoRestartStatus != null) {
	if (shouldResetAutoRestartStatus(previousAutoRestartStatus)) {
	// The connector is not failing now for some time => time to reset the auto-restart status
	LOGGER.infoCr(reconciliation, "Resetting the auto-restart status of connector {} ", connectorName);
	status.autoRestart = null;
	return Future.succeededFuture(status);
	} else {
	// The connector is not failing, but it is not sure yet if it is stable => keep the original status
	status.autoRestart = new AutoRestartStatusBuilder(previousAutoRestartStatus).build();
	return Future.succeededFuture(status);
	}
	} else {
	// No failures and no need to reset the previous auto.restart state => nothing to do
	return Future.succeededFuture(status);
	}
	}
	});
	} else {
	return Future.succeededFuture(status);
	}
	}

Checklist

Please go through this checklist and make sure all applicable tasks have been done

• [ ] Write tests
• [ ] Make sure all tests pass
• Update documentation
• [ ] Check RBAC rights for Kubernetes / OpenShift roles
• [ ] Try your changes from Pod inside your Kubernetes and OpenShift cluster, not just locally
• [ ] Reference relevant issue(s) and close them after merging
• [ ] Update CHANGELOG.md
• [ ] Supply screenshots for visual changes, such as Grafana dashboards

[scholzj]

This seems reasonable. Thanks for the PR!

[rodrigo-molina]

This seems reasonable. Thanks for the PR!

Anytime 😄 !

[ppatierno]

LGTM. Thanks!

[PaulRMellor]

Excellent. Thanks