[jira] [Commented] (FLINK-5158) Handle ZooKeeperCompletedCheckpointStore exceptions in CheckpointCoordinator
[ https://issues.apache.org/jira/browse/FLINK-5158?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15735774#comment-15735774 ] ASF GitHub Bot commented on FLINK-5158: --- Github user tillrohrmann closed the pull request at: https://github.com/apache/flink/pull/2873 > Handle ZooKeeperCompletedCheckpointStore exceptions in CheckpointCoordinator > > > Key: FLINK-5158 > URL: https://issues.apache.org/jira/browse/FLINK-5158 > Project: Flink > Issue Type: Bug > Components: State Backends, Checkpointing >Affects Versions: 1.2.0, 1.1.3 >Reporter: Till Rohrmann >Assignee: Till Rohrmann > Fix For: 1.2.0, 1.1.4 > > > The checkpoint coordinator does not properly handle exceptions when trying to > store completed checkpoints. As a result, completed checkpoints are not > properly cleaned up and even worse, the {{CheckpointCoordinator}} might get > stuck stopping triggering checkpoints. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (FLINK-5158) Handle ZooKeeperCompletedCheckpointStore exceptions in CheckpointCoordinator
[ https://issues.apache.org/jira/browse/FLINK-5158?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15735763#comment-15735763 ] ASF GitHub Bot commented on FLINK-5158: --- Github user asfgit closed the pull request at: https://github.com/apache/flink/pull/2872 > Handle ZooKeeperCompletedCheckpointStore exceptions in CheckpointCoordinator > > > Key: FLINK-5158 > URL: https://issues.apache.org/jira/browse/FLINK-5158 > Project: Flink > Issue Type: Bug > Components: State Backends, Checkpointing >Affects Versions: 1.2.0, 1.1.3 >Reporter: Till Rohrmann >Assignee: Till Rohrmann > Fix For: 1.2.0, 1.1.4 > > > The checkpoint coordinator does not properly handle exceptions when trying to > store completed checkpoints. As a result, completed checkpoints are not > properly cleaned up and even worse, the {{CheckpointCoordinator}} might get > stuck stopping triggering checkpoints. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (FLINK-5158) Handle ZooKeeperCompletedCheckpointStore exceptions in CheckpointCoordinator
[ https://issues.apache.org/jira/browse/FLINK-5158?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15735350#comment-15735350 ] ASF GitHub Bot commented on FLINK-5158: --- Github user tillrohrmann commented on the issue: https://github.com/apache/flink/pull/2872 `ScalaShellITCase` hung in the failing travis build. Merging... > Handle ZooKeeperCompletedCheckpointStore exceptions in CheckpointCoordinator > > > Key: FLINK-5158 > URL: https://issues.apache.org/jira/browse/FLINK-5158 > Project: Flink > Issue Type: Bug > Components: State Backends, Checkpointing >Affects Versions: 1.2.0, 1.1.3 >Reporter: Till Rohrmann >Assignee: Till Rohrmann > Fix For: 1.2.0, 1.1.4 > > > The checkpoint coordinator does not properly handle exceptions when trying to > store completed checkpoints. As a result, completed checkpoints are not > properly cleaned up and even worse, the {{CheckpointCoordinator}} might get > stuck stopping triggering checkpoints. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (FLINK-5158) Handle ZooKeeperCompletedCheckpointStore exceptions in CheckpointCoordinator
[ https://issues.apache.org/jira/browse/FLINK-5158?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15732764#comment-15732764 ] ASF GitHub Bot commented on FLINK-5158: --- Github user tillrohrmann commented on the issue: https://github.com/apache/flink/pull/2872 Forwarding @uce's review from the backport to this PR. Rebasing on the latest master and if Travis gives green light, then I will merge this PR. > Handle ZooKeeperCompletedCheckpointStore exceptions in CheckpointCoordinator > > > Key: FLINK-5158 > URL: https://issues.apache.org/jira/browse/FLINK-5158 > Project: Flink > Issue Type: Bug > Components: State Backends, Checkpointing >Affects Versions: 1.2.0, 1.1.3 >Reporter: Till Rohrmann >Assignee: Till Rohrmann > Fix For: 1.2.0, 1.1.4 > > > The checkpoint coordinator does not properly handle exceptions when trying to > store completed checkpoints. As a result, completed checkpoints are not > properly cleaned up and even worse, the {{CheckpointCoordinator}} might get > stuck stopping triggering checkpoints. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (FLINK-5158) Handle ZooKeeperCompletedCheckpointStore exceptions in CheckpointCoordinator
[ https://issues.apache.org/jira/browse/FLINK-5158?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15712467#comment-15712467 ] ASF GitHub Bot commented on FLINK-5158: --- Github user tillrohrmann commented on the issue: https://github.com/apache/flink/pull/2873 Thanks for the review @uce. I have addressed your comments. > Handle ZooKeeperCompletedCheckpointStore exceptions in CheckpointCoordinator > > > Key: FLINK-5158 > URL: https://issues.apache.org/jira/browse/FLINK-5158 > Project: Flink > Issue Type: Bug > Components: State Backends, Checkpointing >Affects Versions: 1.2.0, 1.1.3 >Reporter: Till Rohrmann >Assignee: Till Rohrmann > Fix For: 1.2.0, 1.1.4 > > > The checkpoint coordinator does not properly handle exceptions when trying to > store completed checkpoints. As a result, completed checkpoints are not > properly cleaned up and even worse, the {{CheckpointCoordinator}} might get > stuck stopping triggering checkpoints. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (FLINK-5158) Handle ZooKeeperCompletedCheckpointStore exceptions in CheckpointCoordinator
[ https://issues.apache.org/jira/browse/FLINK-5158?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15708913#comment-15708913 ] ASF GitHub Bot commented on FLINK-5158: --- Github user uce commented on a diff in the pull request: https://github.com/apache/flink/pull/2873#discussion_r90258638 --- Diff: flink-runtime/src/main/java/org/apache/flink/runtime/checkpoint/CheckpointCoordinator.java --- @@ -731,46 +700,100 @@ public boolean receiveAcknowledgeMessage(AcknowledgeCheckpoint message) throws E discardState(message.getState()); } + + return true; } else if (checkpoint != null) { // this should not happen throw new IllegalStateException( "Received message for discarded but non-removed checkpoint " + checkpointId); } else { + boolean wasPendingCheckpoint; + // message is for an unknown checkpoint, or comes too late (checkpoint disposed) if (recentPendingCheckpoints.contains(checkpointId)) { - isPendingCheckpoint = true; + wasPendingCheckpoint = true; LOG.warn("Received late message for now expired checkpoint attempt {}.", checkpointId); } else { LOG.debug("Received message for an unknown checkpoint {}.", checkpointId); - isPendingCheckpoint = false; + wasPendingCheckpoint = false; } // try to discard the state so that we don't have lingering state lying around discardState(message.getState()); + + return wasPendingCheckpoint; + } + } + } + + private void completePendingCheckpoint(PendingCheckpoint pendingCheckpoint) throws CheckpointException { + final long checkpointId = pendingCheckpoint.getCheckpointId(); + CompletedCheckpoint completedCheckpoint = null; + + try { + completedCheckpoint = pendingCheckpoint.finalizeCheckpoint(); + + completedCheckpointStore.addCheckpoint(completedCheckpoint); + + rememberRecentCheckpointId(checkpointId); + dropSubsumedCheckpoints(completedCheckpoint.getTimestamp()); + + onFullyAcknowledgedCheckpoint(completedCheckpoint); + } catch (Exception exception) { + // abort the current pending checkpoint if it has not been discarded yet + if(!pendingCheckpoint.isDiscarded()) { --- End diff -- missing whitespace after if > Handle ZooKeeperCompletedCheckpointStore exceptions in CheckpointCoordinator > > > Key: FLINK-5158 > URL: https://issues.apache.org/jira/browse/FLINK-5158 > Project: Flink > Issue Type: Bug > Components: State Backends, Checkpointing >Affects Versions: 1.2.0, 1.1.3 >Reporter: Till Rohrmann >Assignee: Till Rohrmann > Fix For: 1.2.0, 1.1.4 > > > The checkpoint coordinator does not properly handle exceptions when trying to > store completed checkpoints. As a result, completed checkpoints are not > properly cleaned up and even worse, the {{CheckpointCoordinator}} might get > stuck stopping triggering checkpoints. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (FLINK-5158) Handle ZooKeeperCompletedCheckpointStore exceptions in CheckpointCoordinator
[ https://issues.apache.org/jira/browse/FLINK-5158?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15708912#comment-15708912 ] ASF GitHub Bot commented on FLINK-5158: --- Github user uce commented on a diff in the pull request: https://github.com/apache/flink/pull/2873#discussion_r90258544 --- Diff: flink-runtime/src/main/java/org/apache/flink/runtime/checkpoint/CheckpointCoordinator.java --- @@ -731,46 +700,100 @@ public boolean receiveAcknowledgeMessage(AcknowledgeCheckpoint message) throws E discardState(message.getState()); } + + return true; } else if (checkpoint != null) { // this should not happen throw new IllegalStateException( "Received message for discarded but non-removed checkpoint " + checkpointId); } else { + boolean wasPendingCheckpoint; + // message is for an unknown checkpoint, or comes too late (checkpoint disposed) if (recentPendingCheckpoints.contains(checkpointId)) { - isPendingCheckpoint = true; + wasPendingCheckpoint = true; LOG.warn("Received late message for now expired checkpoint attempt {}.", checkpointId); } else { LOG.debug("Received message for an unknown checkpoint {}.", checkpointId); - isPendingCheckpoint = false; + wasPendingCheckpoint = false; } // try to discard the state so that we don't have lingering state lying around discardState(message.getState()); + + return wasPendingCheckpoint; + } + } + } + + private void completePendingCheckpoint(PendingCheckpoint pendingCheckpoint) throws CheckpointException { --- End diff -- Missing JavaDocs, maybe add that this needs to be called in checkpoint lock scope > Handle ZooKeeperCompletedCheckpointStore exceptions in CheckpointCoordinator > > > Key: FLINK-5158 > URL: https://issues.apache.org/jira/browse/FLINK-5158 > Project: Flink > Issue Type: Bug > Components: State Backends, Checkpointing >Affects Versions: 1.2.0, 1.1.3 >Reporter: Till Rohrmann >Assignee: Till Rohrmann > Fix For: 1.2.0, 1.1.4 > > > The checkpoint coordinator does not properly handle exceptions when trying to > store completed checkpoints. As a result, completed checkpoints are not > properly cleaned up and even worse, the {{CheckpointCoordinator}} might get > stuck stopping triggering checkpoints. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (FLINK-5158) Handle ZooKeeperCompletedCheckpointStore exceptions in CheckpointCoordinator
[ https://issues.apache.org/jira/browse/FLINK-5158?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15708911#comment-15708911 ] ASF GitHub Bot commented on FLINK-5158: --- Github user uce commented on a diff in the pull request: https://github.com/apache/flink/pull/2873#discussion_r90258796 --- Diff: flink-runtime/src/main/java/org/apache/flink/runtime/checkpoint/CheckpointCoordinator.java --- @@ -731,46 +700,100 @@ public boolean receiveAcknowledgeMessage(AcknowledgeCheckpoint message) throws E discardState(message.getState()); } + + return true; } else if (checkpoint != null) { // this should not happen throw new IllegalStateException( "Received message for discarded but non-removed checkpoint " + checkpointId); } else { + boolean wasPendingCheckpoint; + // message is for an unknown checkpoint, or comes too late (checkpoint disposed) if (recentPendingCheckpoints.contains(checkpointId)) { - isPendingCheckpoint = true; + wasPendingCheckpoint = true; LOG.warn("Received late message for now expired checkpoint attempt {}.", checkpointId); } else { LOG.debug("Received message for an unknown checkpoint {}.", checkpointId); - isPendingCheckpoint = false; + wasPendingCheckpoint = false; } // try to discard the state so that we don't have lingering state lying around discardState(message.getState()); + + return wasPendingCheckpoint; + } + } + } + + private void completePendingCheckpoint(PendingCheckpoint pendingCheckpoint) throws CheckpointException { + final long checkpointId = pendingCheckpoint.getCheckpointId(); + CompletedCheckpoint completedCheckpoint = null; + + try { + completedCheckpoint = pendingCheckpoint.finalizeCheckpoint(); + + completedCheckpointStore.addCheckpoint(completedCheckpoint); + + rememberRecentCheckpointId(checkpointId); + dropSubsumedCheckpoints(completedCheckpoint.getTimestamp()); + + onFullyAcknowledgedCheckpoint(completedCheckpoint); + } catch (Exception exception) { + // abort the current pending checkpoint if it has not been discarded yet + if(!pendingCheckpoint.isDiscarded()) { + pendingCheckpoint.discard(userClassLoader); + } + + if (completedCheckpoint != null) { + // we failed to store the completed checkpoint. Let's clean up + final CompletedCheckpoint cc = completedCheckpoint; + + executor.execute(new Runnable() { + @Override + public void run() { + try { + cc.discard(userClassLoader); + } catch (Exception nestedException) { + LOG.warn("Could not properly discard completed checkpoint {}.", cc.getCheckpointID(), nestedException); + } + } + }); } + + throw new CheckpointException("Could not complete the pending checkpoint " + checkpointId + '.', exception); + } finally { + pendingCheckpoints.remove(checkpointId); + + triggerQueuedRequests(); + } + + LOG.info("Completed checkpoint {} (in {} ms).", checkpointId, completedCheckpoint.getDuration()); + + if (LOG.isDebugEnabled()) { --- End diff -- While rebasing you have to make sure to copy the updated string builder here > Handle ZooKeeperCompletedCheckpointStore exceptions in
[jira] [Commented] (FLINK-5158) Handle ZooKeeperCompletedCheckpointStore exceptions in CheckpointCoordinator
[ https://issues.apache.org/jira/browse/FLINK-5158?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15708914#comment-15708914 ] ASF GitHub Bot commented on FLINK-5158: --- Github user uce commented on a diff in the pull request: https://github.com/apache/flink/pull/2873#discussion_r90257889 --- Diff: flink-runtime/src/main/java/org/apache/flink/runtime/checkpoint/CheckpointCoordinator.java --- @@ -651,64 +651,33 @@ public boolean receiveDeclineMessage(DeclineCheckpoint message) { * * @throws Exception If the checkpoint cannot be added to the completed checkpoint store. */ - public boolean receiveAcknowledgeMessage(AcknowledgeCheckpoint message) throws Exception { + public boolean receiveAcknowledgeMessage(AcknowledgeCheckpoint message) throws CheckpointException { if (shutdown || message == null) { return false; } if (!job.equals(message.getJob())) { - LOG.error("Received AcknowledgeCheckpoint message for wrong job: {}", message); + LOG.error("Received wrong AcknowledgeCheckpoint message for job {}: {}", job, message); return false; } final long checkpointId = message.getCheckpointId(); - CompletedCheckpoint completed = null; - PendingCheckpoint checkpoint; - - // Flag indicating whether the ack message was for a known pending - // checkpoint. - boolean isPendingCheckpoint; - synchronized (lock) { // we need to check inside the lock for being shutdown as well, otherwise we // get races and invalid error log messages if (shutdown) { return false; } - checkpoint = pendingCheckpoints.get(checkpointId); + final PendingCheckpoint checkpoint = pendingCheckpoints.get(checkpointId); if (checkpoint != null && !checkpoint.isDiscarded()) { - isPendingCheckpoint = true; switch (checkpoint.acknowledgeTask(message.getTaskExecutionId(), message.getState(), message.getStateSize(), null)) { case SUCCESS: // TODO: Give KV-state to the acknowledgeTask method --- End diff -- Unrelated, but could you remove this TODO since this has been addressed for 1.2, but probably won't be addressed in 1.1 > Handle ZooKeeperCompletedCheckpointStore exceptions in CheckpointCoordinator > > > Key: FLINK-5158 > URL: https://issues.apache.org/jira/browse/FLINK-5158 > Project: Flink > Issue Type: Bug > Components: State Backends, Checkpointing >Affects Versions: 1.2.0, 1.1.3 >Reporter: Till Rohrmann >Assignee: Till Rohrmann > Fix For: 1.2.0, 1.1.4 > > > The checkpoint coordinator does not properly handle exceptions when trying to > store completed checkpoints. As a result, completed checkpoints are not > properly cleaned up and even worse, the {{CheckpointCoordinator}} might get > stuck stopping triggering checkpoints. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (FLINK-5158) Handle ZooKeeperCompletedCheckpointStore exceptions in CheckpointCoordinator
[ https://issues.apache.org/jira/browse/FLINK-5158?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15702214#comment-15702214 ] ASF GitHub Bot commented on FLINK-5158: --- Github user tillrohrmann commented on the issue: https://github.com/apache/flink/pull/2873 @uce if we want to include this into the release 1.1 branch, then we could merge it. The problem has been reported by Twitter. > Handle ZooKeeperCompletedCheckpointStore exceptions in CheckpointCoordinator > > > Key: FLINK-5158 > URL: https://issues.apache.org/jira/browse/FLINK-5158 > Project: Flink > Issue Type: Bug > Components: State Backends, Checkpointing >Affects Versions: 1.2.0, 1.1.3 >Reporter: Till Rohrmann >Assignee: Till Rohrmann > Fix For: 1.2.0, 1.1.4 > > > The checkpoint coordinator does not properly handle exceptions when trying to > store completed checkpoints. As a result, completed checkpoints are not > properly cleaned up and even worse, the {{CheckpointCoordinator}} might get > stuck stopping triggering checkpoints. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (FLINK-5158) Handle ZooKeeperCompletedCheckpointStore exceptions in CheckpointCoordinator
[ https://issues.apache.org/jira/browse/FLINK-5158?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15696175#comment-15696175 ] ASF GitHub Bot commented on FLINK-5158: --- GitHub user tillrohrmann opened a pull request: https://github.com/apache/flink/pull/2872 [FLINK-5158] [ckPtCoord] Handle exceptions from CompletedCheckpointStore in CheckpointCoordinator Handle exceptions from the CompletedCheckpointStore properly in the CheckpointCoordinator. This means that in case of an exception, the completed checkpoint will be properly cleaned up and also the triggering of subsequent checkpoints will be started. You can merge this pull request into a Git repository by running: $ git pull https://github.com/tillrohrmann/flink fixCheckpointCoordinatorExceptionHandling Alternatively you can review and apply these changes as the patch at: https://github.com/apache/flink/pull/2872.patch To close this pull request, make a commit to your master/trunk branch with (at least) the following in the commit message: This closes #2872 commit 063a696b4eb5a259c714818c0b0ba5cc883a596d Author: Till RohrmannDate: 2016-11-24T17:16:28Z [FLINK-5158] [ckPtCoord] Handle exceptions from CompletedCheckpointStore in CheckpointCoordinator Handle exceptions from the CompletedCheckpointStore properly in the CheckpointCoordinator. This means that in case of an exception, the completed checkpoint will be properly cleaned up and also the triggering of subsequent checkpoints will be started. > Handle ZooKeeperCompletedCheckpointStore exceptions in CheckpointCoordinator > > > Key: FLINK-5158 > URL: https://issues.apache.org/jira/browse/FLINK-5158 > Project: Flink > Issue Type: Bug > Components: State Backends, Checkpointing >Affects Versions: 1.2.0, 1.1.3 >Reporter: Till Rohrmann >Assignee: Till Rohrmann > Fix For: 1.2.0, 1.1.4 > > > The checkpoint coordinator does not properly handle exceptions when trying to > store completed checkpoints. As a result, completed checkpoints are not > properly cleaned up and even worse, the {{CheckpointCoordinator}} might get > stuck stopping triggering checkpoints. -- This message was sent by Atlassian JIRA (v6.3.4#6332)