Skip to content

Fix test server time-skipping locks for activities and external signals #3168

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
thejens wants to merge 1 commit into from
+549 −17
Closed

Fix test server time-skipping locks for activities and external signals #3168

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
87 changes: 70 additions & 17 deletions ...t-server/src/main/java/io/temporal/internal/testservice/TestWorkflowMutableStateImpl.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -102,13 +102,16 @@ private interface UpdateProcedure {
private final StartWorkflowExecutionRequest startRequest;
private long nextEventId = 1;
private final Map<Long, StateMachine<ActivityTaskData>> activities = new HashMap<>();
// Retry backoff does not hold an activity attempt lock.
private final Set<Long> activityLocks = new HashSet<>();
private final Map<String, Long> activityById = new HashMap<>();
private final Map<Long, StateMachine<ChildWorkflowData>> childWorkflows = new HashMap<>();
private final Map<Long, StateMachine<NexusOperationData>> nexusOperations = new HashMap<>();
// Tracks cancelRequestedEventId by scheduledEventId, persists after operation removal.
private final Map<Long, Long> nexusCancelRequestedEventIds = new HashMap<>();
private final Map<String, StateMachine<TimerData>> timers = new HashMap<>();
private final Map<String, StateMachine<SignalExternalData>> externalSignals = new HashMap<>();
private final Set<String> externalSignalLocks = new HashSet<>();
private final Map<String, StateMachine<CancelExternalData>> externalCancellations =
new HashMap<>();
private final Map<String, StateMachine<UpdateWorkflowExecutionData>> updates = new HashMap<>();
Expand Down Expand Up @@ -365,8 +368,17 @@ private void update(boolean completeWorkflowTaskUpdate, UpdateProcedure updater,
} catch (Exception e) {
throw Status.INTERNAL.withCause(e).withDescription(e.getMessage()).asRuntimeException();
} finally {
lockHandle.unlock();
lock.unlock();
try {
if (isTerminalState()) {
for (long scheduledEventId : activityLocks) {
timerService.unlockTimeSkipping("closed workflow activity: " + scheduledEventId);
}
activityLocks.clear();
}
} finally {
lockHandle.unlock();
lock.unlock();
}
}
}

Expand Down Expand Up @@ -1030,6 +1042,12 @@ private void processRequestCancelActivityTask(
// command
activity.action(StateMachines.Action.CANCEL, ctx, null, 0);
activities.remove(scheduledEventId);
ctx.onCommit(
historySize -> {
if (activityLocks.remove(scheduledEventId)) {
timerService.unlockTimeSkipping("processRequestCancelActivityTask");
}
});
ctx.setNeedWorkflowTask(true);
}
}
Expand Down Expand Up @@ -1071,6 +1089,7 @@ private void processScheduleActivityTask(
activityScheduleId, TimeoutType.TIMEOUT_TYPE_SCHEDULE_TO_START, attempt),
"Activity ScheduleToStartTimeout");
ctx.lockTimer("processScheduleActivityTask");
ctx.onCommit(historySize -> activityLocks.add(activityScheduleId));
}

/**
Expand Down Expand Up @@ -1243,29 +1262,52 @@ private void processSignalExternalWorkflowExecution(
}
});
ctx.lockTimer("processSignalExternalWorkflowExecution");
ctx.onCommit(historySize -> externalSignalLocks.add(signalId));
}

@Override
public void completeSignalExternalWorkflowExecution(String signalId, String runId) {
update(
ctx -> {
StateMachine<SignalExternalData> signal = getSignal(signalId);
signal.action(Action.COMPLETE, ctx, runId, 0);
scheduleWorkflowTask(ctx);
ctx.unlockTimer("completeSignalExternalWorkflowExecution");
});
lock.lock();
try {
try {
update(
ctx -> {
StateMachine<SignalExternalData> signal = getSignal(signalId);
signal.action(Action.COMPLETE, ctx, runId, 0);
scheduleWorkflowTask(ctx);
});
} finally {
externalSignals.remove(signalId);
if (externalSignalLocks.remove(signalId)) {
timerService.unlockTimeSkipping("completeSignalExternalWorkflowExecution");
}
}
} finally {
lock.unlock();
}
}

@Override
public void failSignalExternalWorkflowExecution(
String signalId, SignalExternalWorkflowExecutionFailedCause cause) {
update(
ctx -> {
StateMachine<SignalExternalData> signal = getSignal(signalId);
signal.action(Action.FAIL, ctx, cause, 0);
scheduleWorkflowTask(ctx);
ctx.unlockTimer("failSignalExternalWorkflowExecution");
});
lock.lock();
try {
try {
update(
ctx -> {
StateMachine<SignalExternalData> signal = getSignal(signalId);
signal.action(Action.FAIL, ctx, cause, 0);
scheduleWorkflowTask(ctx);
});
} finally {
externalSignals.remove(signalId);
if (externalSignalLocks.remove(signalId)) {
timerService.unlockTimeSkipping("failSignalExternalWorkflowExecution");
}
}
} finally {
lock.unlock();
}
}

private StateMachine<SignalExternalData> getSignal(String signalId) {
Expand Down Expand Up @@ -2137,6 +2179,7 @@ public void completeActivityTask(
activity.action(StateMachines.Action.COMPLETE, ctx, request, 0);
removeActivity(scheduledEventId);
scheduleWorkflowTask(ctx);
ctx.onCommit(historySize -> activityLocks.remove(activity.getData().scheduledEventId));
ctx.unlockTimer("completeActivityTask");
});
}
Expand All @@ -2150,6 +2193,7 @@ public void completeActivityTaskById(
activity.action(StateMachines.Action.COMPLETE, ctx, request, 0);
removeActivity(activity.getData().scheduledEventId);
scheduleWorkflowTask(ctx);
ctx.onCommit(historySize -> activityLocks.remove(activity.getData().scheduledEventId));
ctx.unlockTimer("completeActivityTaskById");
});
}
Expand All @@ -2169,6 +2213,7 @@ public void failActivityTask(long scheduledEventId, RespondActivityTaskFailedReq
addActivityRetryTimer(ctx, activity);
}
// Allow time skipping when waiting for retry
ctx.onCommit(historySize -> activityLocks.remove(activity.getData().scheduledEventId));
ctx.unlockTimer("failActivityTask");
});
}
Expand All @@ -2190,7 +2235,11 @@ private void addActivityRetryTimer(RequestContext ctx, StateMachine<ActivityTask
try {
// TODO this lock is getting releases somewhere on the activity completion.
// We should rework it on passing the lockHandle downstream and using it for the release
update(ctx1 -> ctx1.addActivityTask(data.activityTask));
update(
ctx1 -> {
ctx1.addActivityTask(data.activityTask);
ctx1.onCommit(historySize -> activityLocks.add(data.scheduledEventId));
});
} catch (StatusRuntimeException e) {
// NOT_FOUND is expected as timers are not removed
if (e.getStatus().getCode() != Status.Code.NOT_FOUND) {
Expand Down Expand Up @@ -2226,6 +2275,7 @@ public void failActivityTaskById(
} else {
addActivityRetryTimer(ctx, activity);
}
ctx.onCommit(historySize -> activityLocks.remove(activity.getData().scheduledEventId));
ctx.unlockTimer("failActivityTaskById");
});
}
Expand All @@ -2242,6 +2292,7 @@ public void cancelActivityTask(
activity.action(StateMachines.Action.CANCEL, ctx, request, 0);
removeActivity(scheduledEventId);
scheduleWorkflowTask(ctx);
ctx.onCommit(historySize -> activityLocks.remove(activity.getData().scheduledEventId));
ctx.unlockTimer("cancelActivityTask");
});
}
Expand All @@ -2256,6 +2307,7 @@ public void cancelActivityTaskById(
activity.action(StateMachines.Action.CANCEL, ctx, request, 0);
removeActivity(activity.getData().scheduledEventId);
scheduleWorkflowTask(ctx);
ctx.onCommit(historySize -> activityLocks.remove(activity.getData().scheduledEventId));
ctx.unlockTimer("cancelActivityTaskById");
});
}
Expand Down Expand Up @@ -2317,6 +2369,7 @@ private void timeoutActivity(long scheduledEventId, TimeoutType timeoutType, int
}
}
activity.action(StateMachines.Action.TIME_OUT, ctx, timeoutType, 0);
ctx.onCommit(historySize -> activityLocks.remove(scheduledEventId));
if (isTerminalState(activity.getState())) {
removeActivity(scheduledEventId);
scheduleWorkflowTask(ctx);
Expand Down
Loading
Loading