[jira] [Commented] (HIVE-20959) cbo_rp_limit / cbo_limit are flaky - intermittent whitespace difference

["Laszlo Bodor (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HIVE-20959?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16722569#comment-16722569 ] 

Laszlo Bodor commented on HIVE-20959:
-------------------------------------

[~vihangk1] : as I've seen, you checked this on ptest env, did you try the test alone, or in the failing batch of qtests (please find it in the description). It's unlikely that tests could affect each other this way (as they're running in a serial way, right?), but maybe it worth a try...I've no better idea at the moment.

> cbo_rp_limit / cbo_limit are flaky - intermittent whitespace difference
> -----------------------------------------------------------------------
>
>                 Key: HIVE-20959
>                 URL: https://issues.apache.org/jira/browse/HIVE-20959
>             Project: Hive
>          Issue Type: Bug
>            Reporter: Laszlo Bodor
>            Assignee: Laszlo Bodor
>            Priority: Major
>         Attachments: 171-TestMiniLlapLocalCliDriver-dynamic_semijoin_reduction.q-materialized_view_create_rewrite_3.q-vectorization_pushdown.q-and-27-more.txt, TEST-171-TestMiniLlapLocalCliDriver-dynamic_semijoin_reduction.q-materialized_view_create_rewrite_3.q-vectorization_pushdown.q-and-27-more-TEST-org.apache.hadoop.hive.cli.TestMiniLlapLocalCliDriver.xml, diff
>
>
> {code:java}
> Client Execution succeeded but contained differences (error code = 1) after executing cbo_rp_limit.q 
> 11c11
> < 1 4 2 
> --- 
> > 1 4 2
> {code}
> After copying here to jira, cannot found difference, but by checking from original junit xml, there is a whitespace difference in the lines (represented as hex values), between 1 (x31) and 4 (x34). See  [^diff] . Original golden file contains horizontal tab (x09), actual output contains space (x20). 
> The serious thing is that the separator changes to x20, which is wrong, but then in the same line, it changes back to x09.
> {code}
> 20 31 *20* 34 09 32 <- actual
> 20 31 *09* 34 09 32 <- expected
> {code}
> Tried to reproduce it by running the failing batch of qtests locally, but no luck (maybe it's an environment issue)
> {code}
> mvn test -T 1C -Dtest.output.overwrite=true -Pitests,hadoop-2 -pl itests/qtest -pl itests/util -Dtest=TestMiniLlapLocalCliDriver -Dqfile=dynamic_semijoin_reduction.q,materialized_view_create_rewrite_3.q,vectorization_pushdown.q,correlationoptimizer2.q,cbo_gby_empty.q,schema_evol_text_nonvec_part_all_complex_llap_io.q,vectorization_short_regress.q,mapjoin3.q,cross_product_check_1.q,results_cache_quoted_identifiers.q,unionDistinct_3.q,cbo_join.q,correlationoptimizer6.q,union_remove_26.q,cbo_rp_limit.q,convert_decimal64_to_decimal.q,vector_groupby_cube1.q,union2.q,groupby2.q,dynpart_sort_opt_vectorization.q,constraints_optimization.q,exchgpartition2lel.q,retry_failure.q,schema_evol_text_vecrow_part_llap_io.q,sample10.q,vectorized_timestamp_ints_casts.q,auto_sortmerge_join_2.q,bucketizedhiveinputformat.q,cte_mat_2.q,vectorization_8.q
> {code}



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)