hbase-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Apekshit Sharma <a...@cloudera.com>
Subject Re: Flaky tests
Date Fri, 24 Jun 2016 03:39:58 GMT
Oh, this missed my view.
Yeah, these flaky tests are big pain. At the very least, we should file the
jiras for them. I don't think everyone tracks the builds or the dashboard
(yet), but many do track the jiras.
Instead of RMs cracking down on them on every minor/major release, maybe
everyone in community can take up the challenge of trying to fix one
failure every week.
For instance i spent an hour today just trying to fix
TestHRegionWithInMemory flush (couldn't though). I just realized that I
could have used slack group to get some help too.

Anyways, can we at least agree on opening jiras for these failing tests?
More is always welcome.

On Mon, Jun 20, 2016 at 7:02 PM, Mikhail Antonov <antonov@apache.org> wrote:

> A while ago we ran an effort to crack down on broken/flaky tests on
> branch-1, and I guess some of the branch builds are again in a sorry state,
> so writing here to raise awareness (kind of).
> As I'm looking at
> https://builds.apache.org/job/HBase-1.3 and
> https://builds.apache.org/job/HBase-1.4 builds, I'm seeing lot more red
> things than I would like to (and in many cases I can see one out of two
> configurations (1.7 or 1.8) is passing, while another one is failing.
> Here're the few ones which started to appear in the logs often:
> HBASE-16049 TestRowProcessorEndpoint
> HBASE-16051 TestScannerHeartbeatMessages fails on some machines
> HBASE-16075 TestAcidGuarantees
> http://hbase.x10host.com/flaky-tests/ - this is a more broad set, but I'd
> like to start the discussion whether we should be more aggressive in
> dealing with flaky tests (either disable them and file a jira to follow up,
> or cleanup/remove?)
> Thoughts?
> -Mikhail


-- Appy

  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message