MisterRaindrop commented on code in PR #152:
URL: https://github.com/apache/cloudberry-pxf/pull/152#discussion_r3986365861
##########
regression/sql/FDW_HiveSmokeTest.sql:
##########
@@ -1,49 +0,0 @@
--- data prep
--- start_ignore
-{{ GPDB_REMOTE }}\!ssh {{ PGHOST }} mkdir -p {{ TEST_LOCATION }}
-\!mkdir -p {{ TEST_LOCATION }}
-COPY (
- SELECT format('row_%s',i::varchar(255)),
- i,
- i*0.0001,
- i*100000000000,
- CASE WHEN (i%2) = 0 THEN 'true' ELSE 'false' END
- from generate_series(1, 100) s(i)
- ) TO '{{ TEST_LOCATION }}/data.csv'
- WITH (FORMAT 'csv');
-{{ GPDB_REMOTE }}-- if GPDB is remote, will need to scp file down from there
for beeline
-{{ GPDB_REMOTE }}\!scp {{ PGHOST }}:{{ TEST_LOCATION }}/data.csv {{
TEST_LOCATION }}
-{{ HIVE_REMOTE }}-- if hive is remote, will need to scp file up there to load
it in
-{{ HIVE_REMOTE }}\!cat {{ TEST_LOCATION }}/data.csv | ssh {{ HIVE_HOST }}
'mkdir -p {{ TEST_LOCATION }} && cat > {{ TEST_LOCATION }}/data.csv'
-\!{{ BEELINE_CMD }} -f {{ SCRIPT create_hive_smoke_test_database.sql }} -u
'jdbc:hive2://{{ HIVE_HOST }}:10000/default{{ HIVE_PRINCIPAL }}'
-\!{{ BEELINE_CMD }} -f {{ SCRIPT load_small_data.sql }} -u 'jdbc:hive2://{{
HIVE_HOST }}:10000/hive_smoke_test_database_{{ FULL_TESTNAME }}{{
HIVE_PRINCIPAL }}'
--- end_ignore
-
--- FDW test
-CREATE SERVER hive_smoke_test_server
- FOREIGN DATA WRAPPER hive_pxf_fdw
Review Comment:
Could we keep this test until there is an FDW-enabled replacement? The
existing Hive automation tests are not currently exercised in FDW mode: at the
current PR head, the `features_fdw` report shows all 37 `HiveTest` cases as
skipped. Removing this file therefore drops the end-to-end coverage that
creates `hive_pxf_fdw` and reads from a real Hive table. Please port at least
this basic Hive FDW read case to the automation or Testcontainers suite before
deleting it.
##########
regression/sql/FDW_MultiBlockDataSmokeTest.sql:
##########
@@ -1,37 +0,0 @@
--- FDW test
-CREATE SERVER multi_block_data_smoke_test_server
- FOREIGN DATA WRAPPER {{ HCFS_PROTOCOL }}_pxf_fdw
- OPTIONS (config '{{ SERVER_CONFIG }}');
-CREATE USER MAPPING FOR CURRENT_USER SERVER multi_block_data_smoke_test_server;
-CREATE FOREIGN TABLE multi_block_data_smoke_test_foreign_table (
- t1 TEXT,
- a1 INTEGER
- ) SERVER multi_block_data_smoke_test_server
- OPTIONS (resource '{{ HCFS_BUCKET }}{{ TEST_LOCATION }}', format 'csv');
-
--- write to writable table
-INSERT INTO multi_block_data_smoke_test_foreign_table
Review Comment:
Could we port the write side of this test before deleting it? The new
`MultiBlockDataTest` is not equivalent: it copies a generated file into HDFS
and reads it through a `ReadableExternalTable` in the `load` group. It never
performs this 32-million-row insert through FDW or checks the per-segment
distribution. `HdfsWritableTextTest` exercises FDW writes, but only with 100
rows, so the large multi-block write path remains uncovered. Please retain this
test or add an FDW-enabled large-write replacement.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
