Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Original file line number Diff line number Diff line change
Expand Up @@ -12,10 +12,10 @@
import com.hubspot.mesos.protos.MesosTaskState;
import com.hubspot.singularity.ExtendedTaskState;
import com.hubspot.singularity.InvalidSingularityTaskIdException;
import com.hubspot.singularity.LoadBalancerRequestType;
import com.hubspot.singularity.RequestType;
import com.hubspot.singularity.SingularityCreateResult;
import com.hubspot.singularity.SingularityMainModule;
import com.hubspot.singularity.SingularityManagedScheduledExecutorServiceFactory;
import com.hubspot.singularity.SingularityManagedThreadPoolFactory;
import com.hubspot.singularity.SingularityPendingDeploy;
import com.hubspot.singularity.SingularityPendingRequest;
Expand All @@ -24,10 +24,12 @@
import com.hubspot.singularity.SingularityPendingTask;
import com.hubspot.singularity.SingularityRequestWithState;
import com.hubspot.singularity.SingularityTask;
import com.hubspot.singularity.SingularityTaskCleanup;
import com.hubspot.singularity.SingularityTaskHistory;
import com.hubspot.singularity.SingularityTaskHistoryUpdate;
import com.hubspot.singularity.SingularityTaskId;
import com.hubspot.singularity.SingularityTaskStatusHolder;
import com.hubspot.singularity.TaskCleanupType;
import com.hubspot.singularity.async.ExecutorAndQueue;
import com.hubspot.singularity.config.SingularityConfiguration;
import com.hubspot.singularity.data.DeployManager;
Expand Down Expand Up @@ -344,6 +346,29 @@ private StatusUpdateResult unsafeProcessStatusUpdate(
taskIdObj
);
return StatusUpdateResult.KILL_TASK;
} else if (
maybeTaskHistory.isPresent() &&
status.getReason() == Reason.REASON_AGENT_REREGISTERED
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think this is already covered by the enclosing isRecoveryStatusUpdate if statement

Copy link
Contributor Author

@pschoenfelder pschoenfelder Sep 10, 2020

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I don't see anything specific to re-registration in that method - will leave as is for now

) {
boolean lbRemovalStarted = taskManager
.getLoadBalancerState(taskIdObj, LoadBalancerRequestType.REMOVE)
.isPresent();
if (lbRemovalStarted) {
taskManager.createTaskCleanup(
new SingularityTaskCleanup(
Optional.empty(),
TaskCleanupType.DECOMISSIONING,
System.currentTimeMillis(),
taskIdObj,
Optional.of(
"Agent re-registered after load balancer removal started. Task cannot be reactivated."
),
Optional.empty(),
Optional.empty()
)
);
return StatusUpdateResult.DONE;
}
}
boolean reactivated = taskManager.reactivateTask(
taskIdObj,
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -16,6 +16,7 @@
import com.hubspot.singularity.DeployState;
import com.hubspot.singularity.ExtendedTaskState;
import com.hubspot.singularity.LoadBalancerRequestType;
import com.hubspot.singularity.LoadBalancerRequestType.LoadBalancerRequestId;
import com.hubspot.singularity.MachineState;
import com.hubspot.singularity.RequestCleanupType;
import com.hubspot.singularity.RequestState;
Expand All @@ -29,6 +30,7 @@
import com.hubspot.singularity.SingularityDeployStatistics;
import com.hubspot.singularity.SingularityKilledTaskIdRecord;
import com.hubspot.singularity.SingularityLoadBalancerUpdate;
import com.hubspot.singularity.SingularityLoadBalancerUpdate.LoadBalancerMethod;
import com.hubspot.singularity.SingularityPendingRequest;
import com.hubspot.singularity.SingularityPendingRequest.PendingType;
import com.hubspot.singularity.SingularityPendingTask;
Expand Down Expand Up @@ -4229,6 +4231,68 @@ public void testRecoveredTask() {
Assertions.assertEquals(1, requestManager.getSizeOfPendingQueue());
}

@Test
public void testRecoveredTaskIsCleanedIfLoadBalancerRemoveIsStarted() {
// set up the slave first
sms
.resourceOffers(
Arrays.asList(createOffer(1, 129, 1025, "slave1", "host1", Optional.of("rack1")))
)
.join();

initRequest();
initFirstDeploy();
SingularityTask task = launchTask(request, firstDeploy, 1, TaskState.TASK_RUNNING);

Assertions.assertEquals(1, taskManager.getNumActiveTasks());
TaskStatus lost = TaskStatus
.newBuilder()
.setTaskId(MesosProtosUtils.toTaskId(task.getMesosTask().getTaskId()))
.setAgentId(MesosProtosUtils.toAgentId(task.getAgentId()))
.setReason(Reason.REASON_AGENT_REMOVED)
.setMessage("health check timed out")
.setState(TaskState.TASK_LOST)
.build();

sms.statusUpdate(lost).join();

Assertions.assertEquals(0, taskManager.getNumActiveTasks());
SingularityTaskId taskId = task.getTaskId();
Assertions.assertTrue(taskManager.getTaskHistory(taskId).isPresent());

taskManager.saveLoadBalancerState(
taskId,
LoadBalancerRequestType.REMOVE,
new SingularityLoadBalancerUpdate(
BaragonRequestState.UNKNOWN,
new LoadBalancerRequestId(
taskId.getId(),
LoadBalancerRequestType.REMOVE,
Optional.<Integer>empty()
),
Optional.empty(),
System.currentTimeMillis(),
LoadBalancerMethod.DELETE,
Optional.empty()
)
);

TaskStatus recovered = TaskStatus
.newBuilder()
.setTaskId(MesosProtosUtils.toTaskId(task.getMesosTask().getTaskId()))
.setAgentId(MesosProtosUtils.toAgentId(task.getAgentId()))
.setReason(Reason.REASON_AGENT_REREGISTERED)
.setMessage("agent reregistered")
.setState(TaskState.TASK_RUNNING)
.build();

sms.statusUpdate(recovered).join();

Assertions.assertEquals(0, taskManager.getNumActiveTasks());
Assertions.assertEquals(1, taskManager.getNumCleanupTasks());
Assertions.assertEquals(1, requestManager.getSizeOfPendingQueue());
}

@Test
public void itRetriesLostShortRunningRequests() {
runTest(RequestType.ON_DEMAND, Reason.REASON_AGENT_RESTARTED, true);
Expand Down