You are viewing a plain text version of this content. The canonical link for it is here.
Posted to reviews@helix.apache.org by GitBox <gi...@apache.org> on 2020/09/29 01:19:00 UTC
[GitHub] [helix] alirezazamani opened a new pull request #1422: Fix currentState not being removed when pending message existed
alirezazamani opened a new pull request #1422:
URL: https://github.com/apache/helix/pull/1422
### Issues
- [X] My PR addresses the following Helix issues and references them in the PR description:
Fixes #1421
### Description
- [x] Here are some details about my PR, including screenshots of any UI changes:
With this PR, we make sure that the controller sends DROPPED messages for the jobs that have just finished and their tasks have pending messages.
### Tests
- [x] The following tests are written for this issue:
TestJobCurrentStateDroppedAfterCompletion
- [x] The following is the result of the "mvn test" command on the appropriate module:
```
[INFO] Results:
[INFO]
[ERROR] Failures:
[ERROR] TestDisableCustomCodeRunner.test:233 expected:<false> but was:<true>
[INFO]
[ERROR] Tests run: 1210, Failures: 1, Errors: 0, Skipped: 0
[INFO]
[INFO] ------------------------------------------------------------------------
[INFO] BUILD FAILURE
[INFO] ------------------------------------------------------------------------
[INFO] Total time: 01:11 h
[INFO] Finished at: 2020-09-28T17:49:41-07:00
[INFO] ------------------------------------------------------------------------
```
The failed test individually:
mvn test -Dtest="TestDisableCustomCodeRunner"
```
[INFO] Tests run: 1, Failures: 0, Errors: 0, Skipped: 0, Time elapsed: 8.996 s - in org.apache.helix.integration.TestDisableCustomCodeRunner
[INFO]
[INFO] Results:
[INFO]
[INFO] Tests run: 1, Failures: 0, Errors: 0, Skipped: 0
[INFO]
[INFO] ------------------------------------------------------------------------
[INFO] BUILD SUCCESS
[INFO] ------------------------------------------------------------------------
[INFO] Total time: 14.768 s
[INFO] Finished at: 2020-09-28T18:17:04-07:00
[INFO] ------------------------------------------------------------------------
```
### Commits
- My commits all reference appropriate Apache Helix GitHub issues in their subject lines. In addition, my commits follow the guidelines from "[How to write a good git commit message](http://chris.beams.io/posts/git-commit/)":
1. Subject is separated from body by a blank line
1. Subject is limited to 50 characters (not including Jira issue reference)
1. Subject does not end with a period
1. Subject uses the imperative mood ("add", not "adding")
1. Body wraps at 72 characters
1. Body explains "what" and "why", not "how"
### Code Quality
- My diff has been formatted using helix-style.xml
(helix-style-intellij.xml if IntelliJ IDE is used)
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@helix.apache.org
For additional commands, e-mail: reviews-help@helix.apache.org
[GitHub] [helix] jiajunwang commented on a change in pull request #1422: Fix currentState not being removed when pending message existed
Posted by GitBox <gi...@apache.org>.
jiajunwang commented on a change in pull request #1422:
URL: https://github.com/apache/helix/pull/1422#discussion_r498546089
##########
File path: helix-core/src/test/java/org/apache/helix/integration/task/TestJobCurrentStateDroppedAfterCompletion.java
##########
@@ -0,0 +1,102 @@
+package org.apache.helix.integration.task;
+
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied. See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+import org.apache.helix.AccessOption;
+import org.apache.helix.TestHelper;
+import org.apache.helix.ZkTestHelper;
+import org.apache.helix.task.JobConfig;
+import org.apache.helix.task.JobQueue;
+import org.apache.helix.task.TaskState;
+import org.apache.helix.task.TaskUtil;
+import org.apache.helix.zookeeper.datamodel.ZNRecord;
+import org.apache.helix.zookeeper.impl.client.ZkClient;
+import org.apache.zookeeper.data.Stat;
+import org.testng.Assert;
+import org.testng.annotations.AfterClass;
+import org.testng.annotations.BeforeClass;
+import org.testng.annotations.Test;
+
+import com.google.common.collect.ImmutableMap;
+
+public class TestJobCurrentStateDroppedAfterCompletion extends TaskTestBase {
+
+ @BeforeClass
+ public void beforeClass() throws Exception {
+ _numPartitions = 1;
+ _numNodes = 1;
+ super.beforeClass();
+ }
+
+ @AfterClass
+ public void afterClass() throws Exception {
+ super.afterClass();
+ }
+
+ @Test
+ public void testJobCurrentStateDroppedAfterCompletion() throws Exception {
Review comment:
nit, is it possible to add this test to some existing test class? Recreating the whole test environment is expensive. And I feel we are adding more test classes than necessary. It will soon become unmanageable.
##########
File path: helix-core/src/main/java/org/apache/helix/task/AbstractTaskDispatcher.java
##########
@@ -122,30 +122,37 @@ public void updatePreviousAssignedTasksStatus(
Set<Integer> donePartitions = new TreeSet<>();
for (int pId : pSet) {
final String pName = pName(jobResource, pId);
- TaskPartitionState currState = updateJobContextAndGetTaskCurrentState(currStateOutput,
+ TaskPartitionState currState = getTaskCurrentState(currStateOutput,
jobResource, pId, pName, instance, jobCtx, jobTgtState);
- if (!instance.equals(jobCtx.getAssignedParticipant(pId))) {
- LOG.warn(
- "Instance {} does not match the assigned participant for pId {} in the job context. Skipping task scheduling.",
- instance, pId);
- continue;
- }
-
// Check for pending state transitions on this (partition, instance). If there is a pending
// state transition, we prioritize this pending state transition and set the assignment from
// this pending state transition, essentially "waiting" until this pending message clears
+ // If there is a pending message, we should not continue to update the context because from
Review comment:
This means even the pending message target state matches the current state, we won't proceed? I know it is safer. But since it rolls back this enhancement, it might increase latency. Or does the removed check that the target state matches the current state will introduce any bug?
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@helix.apache.org
For additional commands, e-mail: reviews-help@helix.apache.org
[GitHub] [helix] alirezazamani commented on a change in pull request #1422: Fix currentState not being removed when pending message existed
Posted by GitBox <gi...@apache.org>.
alirezazamani commented on a change in pull request #1422:
URL: https://github.com/apache/helix/pull/1422#discussion_r500409535
##########
File path: helix-core/src/main/java/org/apache/helix/task/AbstractTaskDispatcher.java
##########
@@ -122,30 +122,37 @@ public void updatePreviousAssignedTasksStatus(
Set<Integer> donePartitions = new TreeSet<>();
for (int pId : pSet) {
final String pName = pName(jobResource, pId);
- TaskPartitionState currState = updateJobContextAndGetTaskCurrentState(currStateOutput,
+ TaskPartitionState currState = getTaskCurrentState(currStateOutput,
jobResource, pId, pName, instance, jobCtx, jobTgtState);
- if (!instance.equals(jobCtx.getAssignedParticipant(pId))) {
- LOG.warn(
- "Instance {} does not match the assigned participant for pId {} in the job context. Skipping task scheduling.",
- instance, pId);
- continue;
- }
-
// Check for pending state transitions on this (partition, instance). If there is a pending
// state transition, we prioritize this pending state transition and set the assignment from
// this pending state transition, essentially "waiting" until this pending message clears
+ // If there is a pending message, we should not continue to update the context because from
+ // controller prospective, state transition has not been completed yet if pending message
+ // still existed.
+ // If context gets updated here, controller might remove the job from RunTimeJobDAG which
+ // can cause the task's CurrentState not being removed when there is a pending message for
+ // that task.
Message pendingMessage =
currStateOutput.getPendingMessage(jobResource, new Partition(pName), instance);
- if (pendingMessage != null && !pendingMessage.getToState().equals(currState.name())) {
- // If there is a pending message whose destination state is different from the current
- // state, just make the same assignment as the pending message. This is essentially
- // "waiting" until this state transition is complete
+ if (pendingMessage != null) {
processTaskWithPendingMessage(pId, pName, instance, pendingMessage, jobState, currState,
paMap, assignedPartitions);
continue;
}
Review comment:
It is related to legacy code. Previously, when a job is completed, failed, or timed out, IS would be deleted by the controller. In the next pipeline, since the Ideal state does not exist but the current state exists, we go to legacy code and drop the remaining current states.
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@helix.apache.org
For additional commands, e-mail: reviews-help@helix.apache.org
[GitHub] [helix] alirezazamani commented on a change in pull request #1422: Fix currentState not being removed when pending message existed
Posted by GitBox <gi...@apache.org>.
alirezazamani commented on a change in pull request #1422:
URL: https://github.com/apache/helix/pull/1422#discussion_r500415345
##########
File path: helix-core/src/main/java/org/apache/helix/task/AbstractTaskDispatcher.java
##########
@@ -122,30 +122,37 @@ public void updatePreviousAssignedTasksStatus(
Set<Integer> donePartitions = new TreeSet<>();
for (int pId : pSet) {
final String pName = pName(jobResource, pId);
- TaskPartitionState currState = updateJobContextAndGetTaskCurrentState(currStateOutput,
+ TaskPartitionState currState = getTaskCurrentState(currStateOutput,
jobResource, pId, pName, instance, jobCtx, jobTgtState);
- if (!instance.equals(jobCtx.getAssignedParticipant(pId))) {
- LOG.warn(
- "Instance {} does not match the assigned participant for pId {} in the job context. Skipping task scheduling.",
- instance, pId);
- continue;
- }
-
// Check for pending state transitions on this (partition, instance). If there is a pending
// state transition, we prioritize this pending state transition and set the assignment from
// this pending state transition, essentially "waiting" until this pending message clears
+ // If there is a pending message, we should not continue to update the context because from
+ // controller prospective, state transition has not been completed yet if pending message
+ // still existed.
+ // If context gets updated here, controller might remove the job from RunTimeJobDAG which
+ // can cause the task's CurrentState not being removed when there is a pending message for
+ // that task.
Message pendingMessage =
currStateOutput.getPendingMessage(jobResource, new Partition(pName), instance);
- if (pendingMessage != null && !pendingMessage.getToState().equals(currState.name())) {
Review comment:
It is very complicated to explain:
1- The controller sends RUNNING to COMPLETED message.
2- The participant marks the current state to be COMPLETED but the pending message has not been removed yet.
3- In the next pipeline (pending message toState is COMPLETED, currentState is COMPLETED), this if statement would not be satisfied. Hence, we decide COMPLETED to DROPPED without considering pending messages.
4- The controller marks the job as completed and removes it from the DAG (so we will never consider this job again unless the controller switch to etc).
5- Since the controller sees the pending message, in the message generation phase we do not consider a new decision which is COMPLETED to DROPPED and then CS will never be DROPPED.
The conclusion is that the controller should not consider state transition as a done/completed state transition until the pending message is gone from ZK.
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@helix.apache.org
For additional commands, e-mail: reviews-help@helix.apache.org
[GitHub] [helix] alirezazamani commented on a change in pull request #1422: Fix currentState not being removed when pending message existed
Posted by GitBox <gi...@apache.org>.
alirezazamani commented on a change in pull request #1422:
URL: https://github.com/apache/helix/pull/1422#discussion_r499724813
##########
File path: helix-core/src/main/java/org/apache/helix/task/AbstractTaskDispatcher.java
##########
@@ -122,30 +122,37 @@ public void updatePreviousAssignedTasksStatus(
Set<Integer> donePartitions = new TreeSet<>();
for (int pId : pSet) {
final String pName = pName(jobResource, pId);
- TaskPartitionState currState = updateJobContextAndGetTaskCurrentState(currStateOutput,
+ TaskPartitionState currState = getTaskCurrentState(currStateOutput,
jobResource, pId, pName, instance, jobCtx, jobTgtState);
- if (!instance.equals(jobCtx.getAssignedParticipant(pId))) {
- LOG.warn(
- "Instance {} does not match the assigned participant for pId {} in the job context. Skipping task scheduling.",
- instance, pId);
- continue;
- }
-
// Check for pending state transitions on this (partition, instance). If there is a pending
// state transition, we prioritize this pending state transition and set the assignment from
// this pending state transition, essentially "waiting" until this pending message clears
+ // If there is a pending message, we should not continue to update the context because from
Review comment:
This is more toward the functionality of our scheduling. Without this change, there is a whole in our scheduling decision. Before IS removal change, we were relying on IS to remove the last CS from the participants. However, removing IS revealed this issue. This PR basically handles these cases and helps CS to be cleaned up after the job is being completed. Also, we think about it logically, it makes sense to consider ST not complete when a pending message has not been deleted from the ZK.
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@helix.apache.org
For additional commands, e-mail: reviews-help@helix.apache.org
[GitHub] [helix] NealSun96 commented on a change in pull request #1422: Fix currentState not being removed when pending message existed
Posted by GitBox <gi...@apache.org>.
NealSun96 commented on a change in pull request #1422:
URL: https://github.com/apache/helix/pull/1422#discussion_r500469673
##########
File path: helix-core/src/main/java/org/apache/helix/task/AbstractTaskDispatcher.java
##########
@@ -122,30 +122,37 @@ public void updatePreviousAssignedTasksStatus(
Set<Integer> donePartitions = new TreeSet<>();
for (int pId : pSet) {
final String pName = pName(jobResource, pId);
- TaskPartitionState currState = updateJobContextAndGetTaskCurrentState(currStateOutput,
+ TaskPartitionState currState = getTaskCurrentState(currStateOutput,
jobResource, pId, pName, instance, jobCtx, jobTgtState);
- if (!instance.equals(jobCtx.getAssignedParticipant(pId))) {
- LOG.warn(
- "Instance {} does not match the assigned participant for pId {} in the job context. Skipping task scheduling.",
- instance, pId);
- continue;
- }
-
// Check for pending state transitions on this (partition, instance). If there is a pending
// state transition, we prioritize this pending state transition and set the assignment from
// this pending state transition, essentially "waiting" until this pending message clears
+ // If there is a pending message, we should not continue to update the context because from
+ // controller prospective, state transition has not been completed yet if pending message
+ // still existed.
+ // If context gets updated here, controller might remove the job from RunTimeJobDAG which
+ // can cause the task's CurrentState not being removed when there is a pending message for
+ // that task.
Message pendingMessage =
currStateOutput.getPendingMessage(jobResource, new Partition(pName), instance);
- if (pendingMessage != null && !pendingMessage.getToState().equals(currState.name())) {
Review comment:
Thanks for explaining. So it looks to be a problem with 2 parts:
1. Job context is updated while there are pending messages. This can cause jobs to be removed since it's marked as completed. If DROPPED message isn't correctly sent here, then the DROPPED message will never be sent because the job is deleted;
2. Current state gets updated but pending messages aren't removed (async problem). In that case, the old if statement would not catch the problem and will proceed as if the message doesn't exist.
I suppose 2 is not a problem except when the job is about to be deleted; if the job isn't deleted, the pipeline can simply retry next time.
For 1, I don't understand why the new code in `TaskSchedulingStage` doesn't catch it. Specifically:
```
// Jobs that exist in current states but are missing corresponding JobConfigs or WorkflowConfigs
// or WorkflowContexts need to be cleaned up. Note that restOfResources can only be jobs,
// because workflow resources are created based on Configs only - workflows don't have
// CurrentStates
for (String resourceName : restOfResources.keySet()) {
_workflowDispatcher.processJobForDrop(resourceName, currentStateOutput, output);
}
```
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@helix.apache.org
For additional commands, e-mail: reviews-help@helix.apache.org
[GitHub] [helix] alirezazamani merged pull request #1422: Fix currentState not being removed when pending message existed
Posted by GitBox <gi...@apache.org>.
alirezazamani merged pull request #1422:
URL: https://github.com/apache/helix/pull/1422
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@helix.apache.org
For additional commands, e-mail: reviews-help@helix.apache.org
[GitHub] [helix] NealSun96 commented on a change in pull request #1422: Fix currentState not being removed when pending message existed
Posted by GitBox <gi...@apache.org>.
NealSun96 commented on a change in pull request #1422:
URL: https://github.com/apache/helix/pull/1422#discussion_r499974159
##########
File path: helix-core/src/main/java/org/apache/helix/task/AbstractTaskDispatcher.java
##########
@@ -122,30 +122,37 @@ public void updatePreviousAssignedTasksStatus(
Set<Integer> donePartitions = new TreeSet<>();
for (int pId : pSet) {
final String pName = pName(jobResource, pId);
- TaskPartitionState currState = updateJobContextAndGetTaskCurrentState(currStateOutput,
+ TaskPartitionState currState = getTaskCurrentState(currStateOutput,
jobResource, pId, pName, instance, jobCtx, jobTgtState);
- if (!instance.equals(jobCtx.getAssignedParticipant(pId))) {
- LOG.warn(
- "Instance {} does not match the assigned participant for pId {} in the job context. Skipping task scheduling.",
- instance, pId);
- continue;
- }
-
// Check for pending state transitions on this (partition, instance). If there is a pending
// state transition, we prioritize this pending state transition and set the assignment from
// this pending state transition, essentially "waiting" until this pending message clears
+ // If there is a pending message, we should not continue to update the context because from
+ // controller prospective, state transition has not been completed yet if pending message
+ // still existed.
+ // If context gets updated here, controller might remove the job from RunTimeJobDAG which
+ // can cause the task's CurrentState not being removed when there is a pending message for
+ // that task.
Message pendingMessage =
currStateOutput.getPendingMessage(jobResource, new Partition(pName), instance);
- if (pendingMessage != null && !pendingMessage.getToState().equals(currState.name())) {
- // If there is a pending message whose destination state is different from the current
- // state, just make the same assignment as the pending message. This is essentially
- // "waiting" until this state transition is complete
+ if (pendingMessage != null) {
processTaskWithPendingMessage(pId, pName, instance, pendingMessage, jobState, currState,
paMap, assignedPartitions);
continue;
}
Review comment:
General comment: how did the legacy pipeline help masking this problem before? Or, was it not related to the legacy pipeline, but related to how we delete CurrentStates? The legacy pipeline logic processes the inflight jobs again, but it shouldn't make a difference?
##########
File path: helix-core/src/main/java/org/apache/helix/task/AbstractTaskDispatcher.java
##########
@@ -122,30 +122,37 @@ public void updatePreviousAssignedTasksStatus(
Set<Integer> donePartitions = new TreeSet<>();
for (int pId : pSet) {
final String pName = pName(jobResource, pId);
- TaskPartitionState currState = updateJobContextAndGetTaskCurrentState(currStateOutput,
+ TaskPartitionState currState = getTaskCurrentState(currStateOutput,
jobResource, pId, pName, instance, jobCtx, jobTgtState);
- if (!instance.equals(jobCtx.getAssignedParticipant(pId))) {
- LOG.warn(
- "Instance {} does not match the assigned participant for pId {} in the job context. Skipping task scheduling.",
- instance, pId);
- continue;
- }
-
// Check for pending state transitions on this (partition, instance). If there is a pending
// state transition, we prioritize this pending state transition and set the assignment from
// this pending state transition, essentially "waiting" until this pending message clears
+ // If there is a pending message, we should not continue to update the context because from
+ // controller prospective, state transition has not been completed yet if pending message
+ // still existed.
+ // If context gets updated here, controller might remove the job from RunTimeJobDAG which
+ // can cause the task's CurrentState not being removed when there is a pending message for
+ // that task.
Message pendingMessage =
currStateOutput.getPendingMessage(jobResource, new Partition(pName), instance);
- if (pendingMessage != null && !pendingMessage.getToState().equals(currState.name())) {
Review comment:
It looks like it's just about the order of jobContext updating vs processing pending messages, if I understood you correctly. If that's the case, is there a reason why we removed the second section of the if statement here, as @jiajunwang mentioned?
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@helix.apache.org
For additional commands, e-mail: reviews-help@helix.apache.org
[GitHub] [helix] alirezazamani commented on pull request #1422: Fix currentState not being removed when pending message existed
Posted by GitBox <gi...@apache.org>.
alirezazamani commented on pull request #1422:
URL: https://github.com/apache/helix/pull/1422#issuecomment-704456642
This PR is ready to be merged.
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@helix.apache.org
For additional commands, e-mail: reviews-help@helix.apache.org
[GitHub] [helix] alirezazamani commented on a change in pull request #1422: Fix currentState not being removed when pending message existed
Posted by GitBox <gi...@apache.org>.
alirezazamani commented on a change in pull request #1422:
URL: https://github.com/apache/helix/pull/1422#discussion_r500478114
##########
File path: helix-core/src/main/java/org/apache/helix/task/AbstractTaskDispatcher.java
##########
@@ -122,30 +122,37 @@ public void updatePreviousAssignedTasksStatus(
Set<Integer> donePartitions = new TreeSet<>();
for (int pId : pSet) {
final String pName = pName(jobResource, pId);
- TaskPartitionState currState = updateJobContextAndGetTaskCurrentState(currStateOutput,
+ TaskPartitionState currState = getTaskCurrentState(currStateOutput,
jobResource, pId, pName, instance, jobCtx, jobTgtState);
- if (!instance.equals(jobCtx.getAssignedParticipant(pId))) {
- LOG.warn(
- "Instance {} does not match the assigned participant for pId {} in the job context. Skipping task scheduling.",
- instance, pId);
- continue;
- }
-
// Check for pending state transitions on this (partition, instance). If there is a pending
// state transition, we prioritize this pending state transition and set the assignment from
// this pending state transition, essentially "waiting" until this pending message clears
+ // If there is a pending message, we should not continue to update the context because from
+ // controller prospective, state transition has not been completed yet if pending message
+ // still existed.
+ // If context gets updated here, controller might remove the job from RunTimeJobDAG which
+ // can cause the task's CurrentState not being removed when there is a pending message for
+ // that task.
Message pendingMessage =
currStateOutput.getPendingMessage(jobResource, new Partition(pName), instance);
- if (pendingMessage != null && !pendingMessage.getToState().equals(currState.name())) {
Review comment:
For 1, it is not related to the deletion of the job. The issue is once you mark the last task of the job in the jobContext as completed, then you remove this job from RunTimeJobDag in jobDispatcher. Then next pipeline you will not even see this job in your RuntimeJobDAG. Now there is no way to drop the current state until JobConfig existed. After JobConfig is removed, then the code you pasted should take care of it.
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@helix.apache.org
For additional commands, e-mail: reviews-help@helix.apache.org
[GitHub] [helix] alirezazamani commented on a change in pull request #1422: Fix currentState not being removed when pending message existed
Posted by GitBox <gi...@apache.org>.
alirezazamani commented on a change in pull request #1422:
URL: https://github.com/apache/helix/pull/1422#discussion_r500478114
##########
File path: helix-core/src/main/java/org/apache/helix/task/AbstractTaskDispatcher.java
##########
@@ -122,30 +122,37 @@ public void updatePreviousAssignedTasksStatus(
Set<Integer> donePartitions = new TreeSet<>();
for (int pId : pSet) {
final String pName = pName(jobResource, pId);
- TaskPartitionState currState = updateJobContextAndGetTaskCurrentState(currStateOutput,
+ TaskPartitionState currState = getTaskCurrentState(currStateOutput,
jobResource, pId, pName, instance, jobCtx, jobTgtState);
- if (!instance.equals(jobCtx.getAssignedParticipant(pId))) {
- LOG.warn(
- "Instance {} does not match the assigned participant for pId {} in the job context. Skipping task scheduling.",
- instance, pId);
- continue;
- }
-
// Check for pending state transitions on this (partition, instance). If there is a pending
// state transition, we prioritize this pending state transition and set the assignment from
// this pending state transition, essentially "waiting" until this pending message clears
+ // If there is a pending message, we should not continue to update the context because from
+ // controller prospective, state transition has not been completed yet if pending message
+ // still existed.
+ // If context gets updated here, controller might remove the job from RunTimeJobDAG which
+ // can cause the task's CurrentState not being removed when there is a pending message for
+ // that task.
Message pendingMessage =
currStateOutput.getPendingMessage(jobResource, new Partition(pName), instance);
- if (pendingMessage != null && !pendingMessage.getToState().equals(currState.name())) {
Review comment:
For 1, it is not related to the deletion of the job. The issue is one you mark the last task of the job in the jobContext as completed, then you remove this job from RunTimeJobDag in jobDispatcher. Then next pipeline you will not even see this job in your RuntimeJobDAG. Now there is no way to drop the current state until JobConfig existed. After JobConfig is removed, then the code you pasted should take care of it.
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@helix.apache.org
For additional commands, e-mail: reviews-help@helix.apache.org
[GitHub] [helix] alirezazamani commented on a change in pull request #1422: Fix currentState not being removed when pending message existed
Posted by GitBox <gi...@apache.org>.
alirezazamani commented on a change in pull request #1422:
URL: https://github.com/apache/helix/pull/1422#discussion_r499716296
##########
File path: helix-core/src/test/java/org/apache/helix/integration/task/TestJobCurrentStateDroppedAfterCompletion.java
##########
@@ -0,0 +1,102 @@
+package org.apache.helix.integration.task;
+
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied. See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+import org.apache.helix.AccessOption;
+import org.apache.helix.TestHelper;
+import org.apache.helix.ZkTestHelper;
+import org.apache.helix.task.JobConfig;
+import org.apache.helix.task.JobQueue;
+import org.apache.helix.task.TaskState;
+import org.apache.helix.task.TaskUtil;
+import org.apache.helix.zookeeper.datamodel.ZNRecord;
+import org.apache.helix.zookeeper.impl.client.ZkClient;
+import org.apache.zookeeper.data.Stat;
+import org.testng.Assert;
+import org.testng.annotations.AfterClass;
+import org.testng.annotations.BeforeClass;
+import org.testng.annotations.Test;
+
+import com.google.common.collect.ImmutableMap;
+
+public class TestJobCurrentStateDroppedAfterCompletion extends TaskTestBase {
+
+ @BeforeClass
+ public void beforeClass() throws Exception {
+ _numPartitions = 1;
+ _numNodes = 1;
+ super.beforeClass();
+ }
+
+ @AfterClass
+ public void afterClass() throws Exception {
+ super.afterClass();
+ }
+
+ @Test
+ public void testJobCurrentStateDroppedAfterCompletion() throws Exception {
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.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@helix.apache.org
For additional commands, e-mail: reviews-help@helix.apache.org
[GitHub] [helix] alirezazamani commented on a change in pull request #1422: Fix currentState not being removed when pending message existed
Posted by GitBox <gi...@apache.org>.
alirezazamani commented on a change in pull request #1422:
URL: https://github.com/apache/helix/pull/1422#discussion_r500415345
##########
File path: helix-core/src/main/java/org/apache/helix/task/AbstractTaskDispatcher.java
##########
@@ -122,30 +122,37 @@ public void updatePreviousAssignedTasksStatus(
Set<Integer> donePartitions = new TreeSet<>();
for (int pId : pSet) {
final String pName = pName(jobResource, pId);
- TaskPartitionState currState = updateJobContextAndGetTaskCurrentState(currStateOutput,
+ TaskPartitionState currState = getTaskCurrentState(currStateOutput,
jobResource, pId, pName, instance, jobCtx, jobTgtState);
- if (!instance.equals(jobCtx.getAssignedParticipant(pId))) {
- LOG.warn(
- "Instance {} does not match the assigned participant for pId {} in the job context. Skipping task scheduling.",
- instance, pId);
- continue;
- }
-
// Check for pending state transitions on this (partition, instance). If there is a pending
// state transition, we prioritize this pending state transition and set the assignment from
// this pending state transition, essentially "waiting" until this pending message clears
+ // If there is a pending message, we should not continue to update the context because from
+ // controller prospective, state transition has not been completed yet if pending message
+ // still existed.
+ // If context gets updated here, controller might remove the job from RunTimeJobDAG which
+ // can cause the task's CurrentState not being removed when there is a pending message for
+ // that task.
Message pendingMessage =
currStateOutput.getPendingMessage(jobResource, new Partition(pName), instance);
- if (pendingMessage != null && !pendingMessage.getToState().equals(currState.name())) {
Review comment:
It is very complicated to explain:
1- The controller sends RUNNINg to COMPLETED message.
2- The participant marks the current state to be COMPLETED but the pending message has not been removed yet.
3- In the next pipeline (pending message toState is COMPLETED, currentState is COMPLETED), this if statement would not be satisfied. Hence, we decide COMPLETED to DROPPED without considering pending messages.
4- The controller marks the job as completed and removes it from the DAG (so we will never consider this job again unless the controller switch to etc).
5- Since the controller sees the pending message, in the message generation phase we do not consider a new decision which is COMPLETED to DROPPED and then CS will never be DROPPED.
The conclusion is that the controller should not consider state transition as done state transition until the pending message is gone from ZK.
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@helix.apache.org
For additional commands, e-mail: reviews-help@helix.apache.org
[GitHub] [helix] alirezazamani edited a comment on pull request #1422: Fix currentState not being removed when pending message existed
Posted by GitBox <gi...@apache.org>.
alirezazamani edited a comment on pull request #1422:
URL: https://github.com/apache/helix/pull/1422#issuecomment-704456642
This PR is ready to be merged.
Final commit message:
Fix currentState not being removed when pending message existed
This commit makes sure that the controller drops the current state of
the jobs that have just finished and have pending messages.
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@helix.apache.org
For additional commands, e-mail: reviews-help@helix.apache.org
[GitHub] [helix] alirezazamani commented on a change in pull request #1422: Fix currentState not being removed when pending message existed
Posted by GitBox <gi...@apache.org>.
alirezazamani commented on a change in pull request #1422:
URL: https://github.com/apache/helix/pull/1422#discussion_r500415345
##########
File path: helix-core/src/main/java/org/apache/helix/task/AbstractTaskDispatcher.java
##########
@@ -122,30 +122,37 @@ public void updatePreviousAssignedTasksStatus(
Set<Integer> donePartitions = new TreeSet<>();
for (int pId : pSet) {
final String pName = pName(jobResource, pId);
- TaskPartitionState currState = updateJobContextAndGetTaskCurrentState(currStateOutput,
+ TaskPartitionState currState = getTaskCurrentState(currStateOutput,
jobResource, pId, pName, instance, jobCtx, jobTgtState);
- if (!instance.equals(jobCtx.getAssignedParticipant(pId))) {
- LOG.warn(
- "Instance {} does not match the assigned participant for pId {} in the job context. Skipping task scheduling.",
- instance, pId);
- continue;
- }
-
// Check for pending state transitions on this (partition, instance). If there is a pending
// state transition, we prioritize this pending state transition and set the assignment from
// this pending state transition, essentially "waiting" until this pending message clears
+ // If there is a pending message, we should not continue to update the context because from
+ // controller prospective, state transition has not been completed yet if pending message
+ // still existed.
+ // If context gets updated here, controller might remove the job from RunTimeJobDAG which
+ // can cause the task's CurrentState not being removed when there is a pending message for
+ // that task.
Message pendingMessage =
currStateOutput.getPendingMessage(jobResource, new Partition(pName), instance);
- if (pendingMessage != null && !pendingMessage.getToState().equals(currState.name())) {
Review comment:
It is very complicated to explain:
1- The controller sends RUNNING to COMPLETED message.
2- The participant marks the current state to be COMPLETED but the pending message has not been removed yet.
3- In the next pipeline (pending message toState is COMPLETED, currentState is COMPLETED), this if statement would not be satisfied. Hence, we decide COMPLETED to DROPPED without considering pending messages.
4- The controller marks the job as completed and removes it from the DAG (so we will never consider this job again unless the controller switch to etc).
5- Since the controller sees the pending message, in the message generation phase we do not consider a new decision which is COMPLETED to DROPPED and then CS will never be DROPPED.
The conclusion is that the controller should not consider state transition as done state transition until the pending message is gone from ZK.
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@helix.apache.org
For additional commands, e-mail: reviews-help@helix.apache.org
[GitHub] [helix] NealSun96 commented on a change in pull request #1422: Fix currentState not being removed when pending message existed
Posted by GitBox <gi...@apache.org>.
NealSun96 commented on a change in pull request #1422:
URL: https://github.com/apache/helix/pull/1422#discussion_r500469936
##########
File path: helix-core/src/main/java/org/apache/helix/task/AbstractTaskDispatcher.java
##########
@@ -122,30 +122,37 @@ public void updatePreviousAssignedTasksStatus(
Set<Integer> donePartitions = new TreeSet<>();
for (int pId : pSet) {
final String pName = pName(jobResource, pId);
- TaskPartitionState currState = updateJobContextAndGetTaskCurrentState(currStateOutput,
+ TaskPartitionState currState = getTaskCurrentState(currStateOutput,
jobResource, pId, pName, instance, jobCtx, jobTgtState);
- if (!instance.equals(jobCtx.getAssignedParticipant(pId))) {
- LOG.warn(
- "Instance {} does not match the assigned participant for pId {} in the job context. Skipping task scheduling.",
- instance, pId);
- continue;
- }
-
// Check for pending state transitions on this (partition, instance). If there is a pending
// state transition, we prioritize this pending state transition and set the assignment from
// this pending state transition, essentially "waiting" until this pending message clears
+ // If there is a pending message, we should not continue to update the context because from
+ // controller prospective, state transition has not been completed yet if pending message
+ // still existed.
+ // If context gets updated here, controller might remove the job from RunTimeJobDAG which
+ // can cause the task's CurrentState not being removed when there is a pending message for
+ // that task.
Message pendingMessage =
currStateOutput.getPendingMessage(jobResource, new Partition(pName), instance);
- if (pendingMessage != null && !pendingMessage.getToState().equals(currState.name())) {
- // If there is a pending message whose destination state is different from the current
- // state, just make the same assignment as the pending message. This is essentially
- // "waiting" until this state transition is complete
+ if (pendingMessage != null) {
processTaskWithPendingMessage(pId, pName, instance, pendingMessage, jobState, currState,
paMap, assignedPartitions);
continue;
}
Review comment:
Feel free to close this. My questions are combined at the previous comment.
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@helix.apache.org
For additional commands, e-mail: reviews-help@helix.apache.org