Hi, Here are some updates regarding the ongoing pulsar test instabilities.
The following issues are believed to be caused by PulsarEmbeddedRuntime or mockRuntime, and currently, we have a PR <https://github.com/apache/flink/pull/18838/commits> https://github.com/apache/flink/pull/18838/commits to use PulsarContainerRuntime. Suspected causes are - FLINK-25815: mock pulsar runtime, changed to test containers. - FLINK-26177 - FLINK-26192 - FLINK-26237 - FLINK-26240 - FLINK-26238 The following two tickets are tracked by the corresponding PR - FLINK-26241: Divide zero, a bug in test code. Tracked by <https://github.com/apache/flink/pull/18836> https://github.com/apache/flink/pull/18836 - FLINK-26210: Pulsar e2e tests with jaxb-api. Tracked by <https://github.com/apache/flink/pull/18815> https://github.com/apache/flink/pull/18815 Cheers. On Fri, Feb 18, 2022 at 4:34 PM Yufei Zhang <affei...@gmail.com> wrote: > Hi, > > Sorry for any inconvenience and late reply, after some investigation, the > test instabilities are believed to be caused by the newly introduced > PulsarEmbedRuntime, and an alternative to it is to use the > PulsarContainerRuntime > which runs in testcontainer. A pr is being submitted to resolve the issue > by using PulsarContainerRuntime, and subsequent debugging is ongoing. > Thanks for your update, and we sincerely apologize for the situation. > > > > Cheers. > > On Fri, Feb 18, 2022 at 4:06 PM Chesnay Schepler <ches...@apache.org> > wrote: > >> Given these instabilities we might also want to consider excluding it >> from the 1.15 release, because it is clearly not in a reasonable stable >> state before the feature freeze. >> >> On 18/02/2022 08:58, Chesnay Schepler wrote: >> > FYI: I just disabled all Pulsar tests on master in >> > 296c3be7576940d6bd76f4d065a0b70bceb06991. >> > >> > They are currently the main source of test instabilities, and I don't >> > see any real efforts or progress on those being addressed. >> > >> >>
