yrenat commented on code in PR #6046:
URL: https://github.com/apache/texera/pull/6046#discussion_r4056137752
##########
computing-unit-managing-service/src/main/scala/org/apache/texera/service/ComputingUnitManagingService.scala:
##########
@@ -32,9 +32,65 @@ import org.apache.texera.service.resource.{
ComputingUnitManagingResource,
HealthCheckResource
}
+import
org.apache.texera.service.resource.ComputingUnitManagingResource.TerminatedComputingUnitInfo
+import org.slf4j.LoggerFactory
import java.nio.file.Path
+import java.util.concurrent.TimeUnit
class ComputingUnitManagingService extends
Application[ComputingUnitManagingServiceConfiguration] {
+ private val logger =
LoggerFactory.getLogger(classOf[ComputingUnitManagingService])
+
+ private def initSqlServer(): Unit =
+ SqlServer.initConnection(
+ StorageConfig.jdbcUrl,
+ StorageConfig.jdbcUsername,
+ StorageConfig.jdbcPassword
+ )
+
+ private[service] def registerIdleComputingUnitCleanup(
+ environment: Environment,
+ kubernetesComputingUnitEnabled: Boolean =
KubernetesConfig.kubernetesComputingUnitEnabled,
+ idleTimeoutMinutes: Long =
KubernetesConfig.computingUnitIdleTimeoutMinutes,
+ idleCheckIntervalMinutes: Long =
KubernetesConfig.computingUnitIdleCheckIntervalMinutes,
+ terminateIdleComputingUnits: () => List[TerminatedComputingUnitInfo] =
() =>
+ ComputingUnitManagingResource.terminateIdleKubernetesComputingUnits(),
+ logTerminatedUnits: String => Unit = message => logger.info(message),
+ logCleanupFailure: Throwable => Unit = throwable =>
+ logger.warn("Failed to terminate idle Kubernetes computing units",
throwable),
+ scheduleWithFixedDelay: Option[(Runnable, Long, Long, TimeUnit) => Unit]
= None
+ ): Unit = {
+ if (!kubernetesComputingUnitEnabled || idleTimeoutMinutes <= 0) {
+ return
+ }
+ // scheduleWithFixedDelay rejects a non-positive delay, which would abort
service startup.
+ // A misconfigured interval leaves the rest of the service usable, so log
it and skip the sweep.
+ if (idleCheckIntervalMinutes <= 0) {
+ logger.warn(
+ s"Idle Kubernetes computing unit cleanup is disabled: check interval
must be positive " +
+ s"but is $idleCheckIntervalMinutes minute(s)"
+ )
+ return
+ }
+
+ val scheduler = scheduleWithFixedDelay.getOrElse((command, initialDelay,
delay, unit) =>
+ environment.lifecycle
+ .scheduledExecutorService("idle-computing-unit-terminator")
+ .threads(1)
+ .build()
+ .scheduleWithFixedDelay(command, initialDelay, delay, unit)
+ )
+ scheduler(
+ () =>
+ ComputingUnitManagingService.runIdleComputingUnitCleanup(
+ terminateIdleComputingUnits,
+ logTerminatedUnits,
+ logCleanupFailure
+ ),
+ idleCheckIntervalMinutes,
+ idleCheckIntervalMinutes,
Review Comment:
Done
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
