[
https://issues.apache.org/jira/browse/FLINK-18336?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Roman Khachatryan updated FLINK-18336:
--------------------------------------
Description:
To my understanding, failure shouldn't be counted more than once for a single
checkpoint.
However, after a successful checkpoint, all previous failures are cleared.
So this test will currently fail:
{code:java}
TestFailJobCallback callback = new TestFailJobCallback();
CheckpointFailureManager failureManager = new CheckpointFailureManager(2,
callback);
failureManager.handleJobLevelCheckpointException(new
CheckpointException(CHECKPOINT_EXPIRED), 1L);
failureManager.handleJobLevelCheckpointException(new
CheckpointException(CHECKPOINT_EXPIRED), 2L);
failureManager.handleCheckpointSuccess(2L);
failureManager.handleJobLevelCheckpointException(new
CheckpointException(CHECKPOINT_EXPIRED), 3L);
failureManager.handleJobLevelCheckpointException(new
CheckpointException(CHECKPOINT_EXPIRED), 4L);
// shouldn't be counted because 1L has already failed:
failureManager.handleJobLevelCheckpointException(new
CheckpointException(CHECKPOINT_EXPIRED), 1L);
assertEquals(0, callback.getInvokeCounter());{code}
> CheckpointFailureManager forgets failed checkpoints after a successful one
> --------------------------------------------------------------------------
>
> Key: FLINK-18336
> URL: https://issues.apache.org/jira/browse/FLINK-18336
> Project: Flink
> Issue Type: Bug
> Components: Runtime / Checkpointing
> Reporter: Roman Khachatryan
> Assignee: Roman Khachatryan
> Priority: Major
> Labels: pull-request-available
>
> To my understanding, failure shouldn't be counted more than once for a single
> checkpoint.
> However, after a successful checkpoint, all previous failures are cleared.
> So this test will currently fail:
>
> {code:java}
> TestFailJobCallback callback = new TestFailJobCallback();
> CheckpointFailureManager failureManager = new CheckpointFailureManager(2,
> callback);
> failureManager.handleJobLevelCheckpointException(new
> CheckpointException(CHECKPOINT_EXPIRED), 1L);
> failureManager.handleJobLevelCheckpointException(new
> CheckpointException(CHECKPOINT_EXPIRED), 2L);
> failureManager.handleCheckpointSuccess(2L);
> failureManager.handleJobLevelCheckpointException(new
> CheckpointException(CHECKPOINT_EXPIRED), 3L);
> failureManager.handleJobLevelCheckpointException(new
> CheckpointException(CHECKPOINT_EXPIRED), 4L);
> // shouldn't be counted because 1L has already failed:
> failureManager.handleJobLevelCheckpointException(new
> CheckpointException(CHECKPOINT_EXPIRED), 1L);
> assertEquals(0, callback.getInvokeCounter());{code}
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
