impala-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "David Knupp (Code Review)" <>
Subject [Impala-ASF-CR] IMPALA-4639: Add pytest option and skip markers for tests that only run locally.
Date Sat, 10 Dec 2016 00:49:25 GMT
David Knupp has posted comments on this change.

Change subject: IMPALA-4639: Add pytest option and skip markers for tests that only run locally.

Patch Set 1:

Commit Message:

PS1, Line 7: Add pytest option and skip markers for tests that only run locally.
> Please add Jira ID.
File tests/common/

PS1, Line 120: class SkipIfRemoteTesting:
             :   """Skip marker for those tests that may not work on remote clusters."""
             :   local_cluster_only = pytest.mark.skipif(pytest.config.option.skip_local_only_tests,
             :       reason="--skip_local_only_tests specified")
> I'm fine with the class name. I'm less enthused on the command line name. W

PS1, Line 123: --skip_local_only_tests
> Could this ever be re-used by some other dev without that command line opti
No, if you don't specify the command line, then this won't have any effect.

PS1, Line 124: hbase_split
> Is this used anywhere?
Thanks, forgot to add that file where it gets used.

PS1, Line 125: HBase is not split on remote cluster the same as on the local cluster.
> This doesn't tell me much, probably because I don't know how splitting work
Hbase splitting is usually non-deterministic, but we enforce a particular split on our local
tests via a setup shell script, and then conform our tests to expect that split. We don't
have a way of doing this on the remote cluster, and it would require a big overhaul to our
existing tests to work without it.
File tests/query_test/

PS1, Line 95: @SkipIfRemoteTesting.local_cluster_only
> From just looking at this it is not clear to me whether this is skipped or 
This will only be skipped if --testing_remote_cluster is supplied on the command line.

PS1, Line 95: @SkipIfRemoteTesting.local_cluster_only
> I think also it's confusing when we have the distinction between a local HD

Line 99:     self.run_test_case('QueryTest/mt-dop-parquet-nested', vector)
> Why isn't this test expected to succeed on a cluster? It seems like it shou
It's a setup problem. We're still working though kinks on getting our entire data load working
across the board on remote clusters. Even though the process completes without any errors,
here are as yet silent failures. I can add a JIRA and reference it here.

To view, visit
To unsubscribe, visit

Gerrit-MessageType: comment
Gerrit-Change-Id: Id4d6e46dc1e64ad20c727ccb19af7a9f3daf917f
Gerrit-PatchSet: 1
Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-Owner: David Knupp <>
Gerrit-Reviewer: Alex Behm <>
Gerrit-Reviewer: David Knupp <>
Gerrit-Reviewer: Lars Volker <>
Gerrit-Reviewer: Michael Brown <>
Gerrit-Reviewer: Tim Armstrong <>
Gerrit-HasComments: Yes

View raw message