Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Use HDP 3.1 by default in product tests #1239

Closed
wants to merge 17 commits into from
Closed

Conversation

findepi
Copy link
Member

@findepi findepi commented Aug 2, 2019

This gives better test coverage in the future.

Tests on HDP 2.6 will be run in "branch builds".

Related to #190, #37, #1218, #1188
Required for #1034

@cla-bot cla-bot bot added the cla-signed label Aug 2, 2019
@findepi findepi added the WIP label Aug 2, 2019
@findepi findepi force-pushed the hdp3 branch 8 times, most recently from 10f9eb2 to c0bdcc3 Compare August 5, 2019 16:15
@findepi findepi force-pushed the hdp3 branch 11 times, most recently from 229181f to cccc825 Compare August 23, 2019 07:39
@findepi findepi force-pushed the hdp3 branch 6 times, most recently from 6576d1b to 08b32a5 Compare September 10, 2019 13:03
Despite
 https://cwiki.apache.org/confluence/display/Hive/LanguageManual+Types#LanguageManualTypes-FloatingPointTypes
saying "Floating point literals are assumed to be DOUBLE. Scientific
notation is not yet supported."

- Hive 1.2 supports scientific notation
- hive 3 interprets decimal literals as decimals
`ALTER TABLE .. ENABLE OFFLINE` was removed in Hive 2.0
Hive 3's metastore returns basic statistics for a table even when we
don't write them.
On HDP 3.1 new schema location is owned by `hive` on HDFS with
`rwxr-xr-x` permissions. `alice` user cannot create a new table under
the schema when HDFS impersonation is enabled.
Previously the `PRODUCT_TESTS_CONFIG=config-hdp3` configurations would
still run on HDP 2.6.
HDP 3.1 gives better test coverage in the future.

Tests on HDP 2.6 will be run in "branch builds".
The test expects all buckets are populated. However, Hive bucketing v1
uses float's bits as a hash, and for small float numbers, the last 5
bits (i.e. module 32 when there are 32 buckets) are usually zeros.

We still got a few buckets (more than one) in
`presto_test_bucketed_by_double_float`, because another hashed column
was double.
There is a second `-Dhive.hadoop2.timeZone` on the same invocation.
Hive 3 is less permissive when it comes to schema evolution.
@findepi
Copy link
Member Author

findepi commented Jan 10, 2020

Obsolete. We run tests with Hive 3.1 and we currently don't run them on Hive 3.1 by default.

@findepi findepi closed this Jan 10, 2020
@findepi findepi deleted the hdp3 branch January 10, 2020 18:02
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Development

Successfully merging this pull request may close these issues.

1 participant