gax-java(windows): Flaky HttpJsonDirectServerStreamingCallableTest.testOnResponseError

[github-actions]

The build has failed : https://github.com/googleapis/gapic-generator-java/actions/runs/4942995036

[mpeddada1]

Stacktrace:

Error:  Tests run: 9, Failures: 1, Errors: 0, Skipped: 0, Time elapsed: 8.529 s <<< FAILURE! - in com.google.api.gax.httpjson.HttpJsonDirectServerStreamingCallableTest
Error:  com.google.api.gax.httpjson.HttpJsonDirectServerStreamingCallableTest.testOnResponseError  Time elapsed: 2.064 s  <<< FAILURE!
value of: await(...)
expected to be true
	at com.google.api.gax.httpjson.HttpJsonDirectServerStreamingCallableTest.testOnResponseError(HttpJsonDirectServerStreamingCallableTest.java:267)

[meltsufin]

@lqiu96 @blakeli0 Can we prioritize fixing this? It's causing a lot of flaky failures.

[lqiu96]

@lqiu96 @blakeli0 Can we prioritize fixing this? It's causing a lot of flaky failures.

Sure. I'll take a look.

[lqiu96]

Having a difficult time reproducing this problem on a Windows cloudtop with Java 11. Ran this test for 40k invocations and all passed. Multiple 10k invocation runs have resulted in all passing tests. Need to take another look. Maybe the nightly job is set up differently?

[blakeli0]

Maybe the nightly job is set up differently?

What is the hardware specification for the cloudtop machine and the Github action machine? I guess the machine Github action uses is much worse than the cloudtop. In which case, I don't think we can do much other than increase the wait time. I gave the test itself a quick review and I don't see anything wrong with the setups.

[lqiu96]

Maybe the nightly job is set up differently?

What is the hardware specification for the cloudtop machine and the Github action machine? I guess the machine Github action uses is much worse than the cloudtop. In which case, I don't think we can do much other than increase the wait time. I gave the test itself a quick review and I don't see anything wrong with the setups.

Yeah that seems very likely. I'll spend a few hours looking into this issue and I'll double the wait time if I'm still stuck.

[blakeli0]

Reopening this issue as it is happening again last night #1939

[suztomo]

Marking this as fixed in #2291. Let's reopen if this recurs.