[
https://issues.apache.org/jira/browse/FLINK-14951?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16991629#comment-16991629
]
Yangze Guo edited comment on FLINK-14951 at 12/9/19 2:30 PM:
-------------------------------------------------------------
Actually, I execute it by run_singe_test.sh. Many tests will touch
"taskmanager.numberOfTaskSlots" configuration, such as
"test_high_parallelism_iterations.sh". After it failure, the cleanup function
will not be called to revert the configuration. Thus, I met this problem.
Since the test case has thread safety issue, I think we should either set the
"taskmanager.numberOfTaskSlots" explicitly to 1 or harden the test case. WDYT?
Agreed that it should not block the 1.10 release. What about setting the fix
version to 1.11?
was (Author: karmagyz):
Actually, I execute it by run_singe_test.sh. Many tests will touch
"taskmanager.numberOfTaskSlots" configuration, such as
"test_high_parallelism_iterations.sh". After it failure, the cleanup function
will not be called to revert the configuration. Thus, I met this problem.
Since the test case has thread safety issue, I think we should either
proactively set the "taskmanager.numberOfTaskSlots" to 1 or harden the test
case. WDYT?
Agreed that it should not block the 1.10 release. What about set the fix
version to 1.11?
> State TTL backend end-to-end test fail when taskManager has multiple slot
> -------------------------------------------------------------------------
>
> Key: FLINK-14951
> URL: https://issues.apache.org/jira/browse/FLINK-14951
> Project: Flink
> Issue Type: Bug
> Components: Runtime / State Backends, Tests
> Environment: centos 7
> java 8
> Reporter: Yangze Guo
> Assignee: Yangze Guo
> Priority: Major
> Labels: pull-request-available
> Fix For: 1.10.0
>
> Time Spent: 10m
> Remaining Estimate: 0h
>
> When I run flink end to end tests, the State TTL backend tests fail. The log
> of TaskManager show below:
> 2019-11-26 20:22:03,837 INFO org.apache.flink.runtime.taskmanager.Task
> - TtlVerifyUpdateFunction -> Sink: PrintFailedVerifications
> (3/3) (23f969ddb3e13fcdd3ba9823f50b0eab) switched from RUNNING to FAILED.
> java.lang.IllegalStateException: Timestamps before and after the update do
> not match.
> at
> org.apache.flink.util.Preconditions.checkState(Preconditions.java:195)
> at
> org.apache.flink.streaming.tests.TtlVerifyUpdateFunction.performUpdate(TtlVerifyUpdateFunction.java:124)
> at
> org.apache.flink.streaming.tests.TtlVerifyUpdateFunction.generateUpdateAndVerificationContext(TtlVerifyUpdateFunction.java:101)
> at
> org.apache.flink.streaming.tests.TtlVerifyUpdateFunction.flatMap(TtlVerifyUpdateFunction.java:88)
> at
> org.apache.flink.streaming.tests.TtlVerifyUpdateFunction.flatMap(TtlVerifyUpdateFunction.java:67)
> at
> org.apache.flink.streaming.api.operators.StreamFlatMap.processElement(StreamFlatMap.java:50)
> at
> org.apache.flink.streaming.runtime.tasks.OneInputStreamTask$StreamTaskNetworkOutput.emitRecord(OneInputStreamTask.java:173)
> at
> org.apache.flink.streaming.runtime.io.StreamTaskNetworkInput.processElement(StreamTaskNetworkInput.java:151)
> at
> org.apache.flink.streaming.runtime.io.StreamTaskNetworkInput.emitNext(StreamTaskNetworkInput.java:128)
> at
> org.apache.flink.streaming.runtime.io.StreamOneInputProcessor.processInput(StreamOneInputProcessor.java:69)
> at
> org.apache.flink.streaming.runtime.tasks.StreamTask.processInput(StreamTask.java:284)
> at
> org.apache.flink.streaming.runtime.tasks.mailbox.MailboxProcessor.runMailboxLoop(MailboxProcessor.java:155)
> at
> org.apache.flink.streaming.runtime.tasks.StreamTask.invoke(StreamTask.java:445)
> at org.apache.flink.runtime.taskmanager.Task.doRun(Task.java:702)
> at org.apache.flink.runtime.taskmanager.Task.run(Task.java:527)
> at java.lang.Thread.run(Thread.java:834)
> It is cause by the MonotonicTTLTimeProvider:freeze and
> MonotonicTTLTimeProvider:unfreezeTime called by multithread when
> taskmanager.numberOfTaskSlots set greater than 1. We could set it to 1 in
> test_stream_state_ttl.sh. That will fix the problem.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
