Support insert into bucketed unpartitioned Hive table

[kokosing]

Support insert into bucketed unpartitioned Hive table

[kokosing]

@electrum @findepi I naively disabled the check state and run some manual tests and it seems to be working. What am I missing? Or what should I also test?

presto:default> show create table b;
         Create Table
-------------------------------
 CREATE TABLE hive.default.b (
    i integer
 )
 WITH (
    bucket_count = 5,
    bucketed_by = ARRAY['i'],
    format = 'ORC',
    sorted_by = ARRAY[]
 )
(1 row)

Query 20190715_140514_00026_vgg79, FINISHED, 1 node
Splits: 1 total, 1 done (100.00%)
0:00 [0 rows, 0B] [0 rows/s, 0B/s]

presto:default> select count(*) from b;
 _col0
-------
    19
(1 row)

Query 20190715_140517_00027_vgg79, FINISHED, 1 node
Splits: 47 total, 47 done (100.00%)
0:01 [19 rows, 4.9KB] [32 rows/s, 8.38KB/s]

presto:default> insert into b select * from b;
INSERT: 19 rows

Query 20190715_140525_00028_vgg79, FINISHED, 1 node
Splits: 48 total, 48 done (100.00%)
0:02 [19 rows, 4.9KB] [9 rows/s, 2.45KB/s]

presto:default> select count(*) from b;
 _col0
-------
    38
(1 row)

Query 20190715_140529_00029_vgg79, FINISHED, 1 node
Splits: 52 total, 52 done (100.00%)
0:01 [38 rows, 6.19KB] [45 rows/s, 7.39KB/s]

[kokosing]

Fixes #1088

[sopel39]

There is another, "similar" check:

                if (bucketNumber.isPresent()) {
                    throw new PrestoException(HIVE_PARTITION_READ_ONLY, "Cannot insert into existing partition of bucketed Hive table: " + partitionName.get());
                }

[kokosing]

There is another, "similar" check:

               if (bucketNumber.isPresent()) {
                   throw new PrestoException(HIVE_PARTITION_READ_ONLY, "Cannot insert into existing > partition of bucketed Hive table: " + partitionName.get());
               }

Yes, this about partitioned and bucketed tables (a bit out of the scope of this PR but strongly related).

[kokosing]

@electrum @dain do you guys have any comments to that? If not I would like to continue on that (adding more tests and handle also partitioned tables)

[dain]

@kokosing I don't have any comments, but I'd like to see @electrum reply.

[kokosing]

@sopel39, @electrum I added more tests, would you like to review?

[sopel39]

I think @electrum also wanted to review it

[sopel39]

remove

[electrum]

This should be test_empty_bucketed_table

[electrum]

test_bucketed_table

[electrum]

Nit: space between table name and (

[kokosing]

[INFO] --- maven-enforcer-plugin:3.0.0-M2:enforce (default) @ presto-hive-hadoop2 ---
[INFO] Ignoring requireUpperBoundDeps in com.google.guava:guava
[INFO] Ignoring requireUpperBoundDeps in org.codehaus.plexus:plexus-utils
[WARNING] Rule 5: org.apache.maven.plugins.enforcer.EnforceBytecodeVersion failed with message:
IOException while reading /home/travis/.m2/repository/software/amazon/ion/ion-java/1.0.2/ion-java-1.0.2.jar

I get the above error for HIVE_TESTS=true and command /mvnw -B -pl presto-hive-hadoop2 test -P test-hive-hadoop2 -Dhive.hadoop2.timeZone=UTC -DHADOOP_USER_NAME=hive -Dhive.hadoop2.metastoreHost=localhost -Dhive.hadoop2.metastorePort=9083 -Dhive.hadoop2.databaseName=default -Dhive.hadoop2.metastoreHost=hadoop-master -Dhive.hadoop2.timeZone=Asia/Kathmandu -Dhive.metastore.thrift.client.socks-proxy=127.0.0.1:1180 -Dhadoop-master-ip=172.18.0.2 in travis. This is passing just fine locally.