[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: org.apache.beam.runners.flink.PortableTimersExecutionTest is very flakey

I actually didn't look at this one. I filed a bunch more adjacent flake bugs. I didn't find your bug but I do see that test flaking at the same time as the others. FWIW here is the list of flakes and sickbayed tests: https://issues.apache.org/jira/issues/?filter=12343195


On Tue, Nov 27, 2018 at 10:25 AM Alex Amato <ajamato@xxxxxxxxxx> wrote:

Did you happen to look into this test? I heard that you may have been looking into this.

On Mon, Nov 26, 2018 at 3:36 PM Maximilian Michels <mxm@xxxxxxxxxx> wrote:
Hi Alex,

Thanks for your help! I'm quite used to debugging concurrent/distributed
problems. But this one is quite tricky, especially with regards to GRPC
threads. I try to provide more information in the following.

There are two observations:

1) The problem is specifically related to how the cleanup is performed
for the EmbeddedEnvironmentFactory. The environment is shutdown when the
SDK Harness exists but the GRPC threads continue to linger for some time
and may stall state processing on the next test.

If you do _not_ close DefaultJobBundleFactory, which happens during
close() or dispose() in the FlinkExecutableStageFunction or
ExecutableStageDoFnOperator respectively, the tests run just fine. I ran
1000 test runs without a single failure.

The EmbeddedEnvironment uses direct channels which are marked
experimental in GRPC. We may have to convert them to regular socket

2) Try setting a conditional breakpoint in GrpcStateService which will
never break, e.g. "false". Set it here:

The tests will never fail. The SDK harness is always shutdown correctly
at the end of the test.


On 26.11.18 19:15, Alex Amato wrote:
> Thanks Maximilian, let me know if you need any help. Usually I debug
> this sort of thing by pausing the IntelliJ debugger to see all the
> different threads which are waiting on various conditions. If you find
> any insights from that, please post them here and we can try to figure
> out the source of the stuckness. Perhaps it may be some concurrency
> issue leading to deadlock?
> On Thu, Nov 22, 2018 at 12:57 PM Maximilian Michels <mxm@xxxxxxxxxx
> <mailto:mxm@xxxxxxxxxx>> wrote:
>     I couldn't fix it thus far. The issue does not seem to be in the Flink
>     Runner but in the way the tests utilizes the EMBEDDED environment to
>     run
>     multiple portable jobs in a row.
>     When it gets stuck it is in RemoteBundle#close and it is independent of
>     the test type (batch and streaming have different implementations).
>     Will give it another look tomorrow.
>     Thanks,
>     Max
>     On 22.11.18 13:07, Maximilian Michels wrote:
>      > Hi Alex,
>      >
>      > The test seems to have gotten flaky after we merged support for
>     portable
>      > timers in Flink's batch mode.
>      >
>      > Looking into this now.
>      >
>      > Thanks,
>      > Max
>      >
>      > On 21.11.18 23:56, Alex Amato wrote:
>      >> Hello, I have noticed
>      >> that org.apache.beam.runners.flink.PortableTimersExecutionTest
>     is very
>      >> flakey, and repro'd this test timeout on the master branch in
>     40/50 runs.
>      >>
>      >> I filed a JIRA issue: BEAM-6111
>      >> <https://issues.apache.org/jira/browse/BEAM-6111>. I was just
>      >> wondering if anyone knew why this may be occurring, and to check if
>      >> anyone else has been experiencing this.
>      >>
>      >> Thanks,
>      >> Alex