[WIP] Tests for debugging Dart and MSQ in Quidem #17861

weishiuntsai · 2025-04-02T01:59:03Z

Add 2 tests for debugging Dart and MSQ in Quidem. Do not merge.

Add 2 tests for debugging Dart and Msq in Quidem.

kgyrtkirk · 2025-04-10T11:42:33Z

downloaded the heapdump - the issue was caused by huge exceptions 5M each - and surefire tried to hold all of them in a map....which have lead to an OOM

the underlying exception is a diff related exception; but it contains the following exception(s):

> WHERE time_floor(__time, 'PT1H') BETWEEN timestamp '2019-08-25 00:00:00' AND timestamp '2019-08-25 06:00:00'": Remote driver error: QueryInterruptedException: Leaks happened, each suppressed exception represents one code path that checked out an object and didn't return it. -> RuntimeException: Leaks happened, each suppressed exception represents one code path that checked out an object and didn't return it.
>       at org.apache.calcite.avatica.Helper.createException(Helper.java:54)
>       at org.apache.calcite.avatica.Helper.createException(Helper.java:41)
[...]
>       Suppressed: org.apache.druid.collections.StupidPool$LeakedException: Originally checked out by thread [Test-runner-processing-pool-msq-worker[00000000-0000-0000-0000-000000000000_2_0]]
>               at org.apache.druid.collections.StupidPool.take(StupidPool.java:162)
>               at org.apache.druid.segment.CompressedPools.getByteBuf(CompressedPools.java:110)
>               at org.apache.druid.segment.data.DecompressingByteBufferObjectStrategy.fromByteBuffer(DecompressingByteBufferObjectStrategy.java:70)
>               at org.apache.druid.segment.data.DecompressingByteBufferObjectStrategy.fromByteBuffer(DecompressingByteBufferObjectStrategy.java:30)
>               at org.apache.druid.segment.data.GenericIndexed$BufferIndexed.get(GenericIndexed.java:598)
>               at org.apache.druid.segment.data.BlockLayoutColumnarLongsSupplier$1.loadBuffer(BlockLayoutColumnarLongsSupplier.java:103)
>               at org.apache.druid.segment.data.BlockLayoutColumnarLongsSupplier$1.get(BlockLayoutColumnarLongsSupplier.java:90)
>               at org.apache.druid.segment.column.LongsColumn.getLongSingleValueRow(LongsColumn.java:77)
>               at org.apache.druid.segment.QueryableIndexCursorHolder.asCursor(QueryableIndexCursorHolder.java:185)
>               at org.apache.druid.msq.querykit.scan.ScanQueryFrameProcessor.runWithSegment(ScanQueryFrameProcessor.java:271)
>               at org.apache.druid.msq.querykit.BaseLeafFrameProcessor.runIncrementally(BaseLeafFrameProcessor.java:88)
>               at org.apache.druid.msq.querykit.scan.ScanQueryFrameProcessor.runIncrementally(ScanQueryFrameProcessor.java:157)
>               at org.apache.druid.msq.counters.CpuTimeAccumulatingFrameProcessor.runIncrementally(CpuTimeAccumulatingFrameProcessor.java:66)
>               at org.apache.druid.frame.processor.FrameProcessors$1FrameProcessorWithBaggage.runIncrementally(FrameProcessors.java:72)
>               at org.apache.druid.frame.processor.FrameProcessorExecutor$1ExecutorRunnable.runProcessorNow(FrameProcessorExecutor.java:239)
>               at org.apache.druid.frame.processor.FrameProcessorExecutor$1ExecutorRunnable.run(FrameProcessorExecutor.java:141)
>               at org.apache.druid.msq.exec.WorkerImpl$2$2.run(WorkerImpl.java:900)
>               ... 3 more
[this repeats around a 40 times]

kgyrtkirk · 2025-04-11T08:12:49Z

running the same query which errors out multiple times leads to the same issue:

tt.iq

!set dartQueryId 00000000-0000-0000-0000-000000000000
!set useApproximateCountDistinct false
!use druidtest://?componentSupplier=DartComponentSupplier&datasets=sql/src/test/quidem/qatests/qaArray/sql&numMergeBuffers=3
!set outputformat mysql


# TESTCASE: test_subquery_with_where TEST_ID: A2_B46_C5 TYPE: NEGATIVE TEST
SELECT a_int
FROM
  (SELECT *
   FROM test_array)
WHERE a_int NOT IN
    (SELECT a_int
     FROM test_array);
ARRAY
!error

# TESTCASE: test_subquery_with_where TEST_ID: A2_B46_C5 TYPE: NEGATIVE TEST
SELECT a_int
FROM
  (SELECT *
   FROM test_array)
WHERE a_int NOT IN
    (SELECT a_int
     FROM test_array);
ARRAY
!error


# TESTCASE: test_subquery_with_where TEST_ID: A2_B46_C5 TYPE: NEGATIVE TEST
SELECT a_int
FROM
  (SELECT *
   FROM test_array)
WHERE a_int NOT IN
    (SELECT a_int
     FROM test_array);
ARRAY
!error

# TESTCASE: test_subquery_with_where TEST_ID: A2_B46_C5 TYPE: NEGATIVE TEST
SELECT a_int
FROM
  (SELECT *
   FROM test_array)
WHERE a_int NOT IN
    (SELECT a_int
     FROM test_array);
ARRAY
!error

# TESTCASE: test_subquery_with_where TEST_ID: A2_B46_C5 TYPE: NEGATIVE TEST
SELECT a_int
FROM
  (SELECT *
   FROM test_array)
WHERE a_int NOT IN
    (SELECT a_int
     FROM test_array);
ARRAY
!error

# TESTCASE: test_subquery_with_where TEST_ID: A2_B46_C5 TYPE: NEGATIVE TEST
SELECT a_int
FROM
  (SELECT *
   FROM test_array)
WHERE a_int NOT IN
    (SELECT a_int
     FROM test_array);
ARRAY
!error

# TESTCASE: test_subquery_with_where TEST_ID: A2_B46_C5 TYPE: NEGATIVE TEST
SELECT a_int
FROM
  (SELECT *
   FROM test_array)
WHERE a_int NOT IN
    (SELECT a_int
     FROM test_array);
ARRAY
!error

# TESTCASE: test_subquery_with_where TEST_ID: A2_B46_C5 TYPE: NEGATIVE TEST
SELECT a_int
FROM
  (SELECT *
   FROM test_array)
WHERE a_int NOT IN
    (SELECT a_int
     FROM test_array);
ARRAY
!error

# TESTCASE: test_subquery_with_where TEST_ID: A2_B46_C5 TYPE: NEGATIVE TEST
SELECT a_int
FROM
  (SELECT *
   FROM test_array)
WHERE a_int NOT IN
    (SELECT a_int
     FROM test_array);
ARRAY
!error

# TESTCASE: test_subquery_with_where TEST_ID: A2_B46_C5 TYPE: NEGATIVE TEST
SELECT a_int
FROM
  (SELECT *
   FROM test_array)
WHERE a_int NOT IN
    (SELECT a_int
     FROM test_array);
ARRAY
!error

# TESTCASE: test_subquery_with_where TEST_ID: A2_B46_C5 TYPE: NEGATIVE TEST
SELECT a_int
FROM
  (SELECT *
   FROM test_array)
WHERE a_int NOT IN
    (SELECT a_int
     FROM test_array);
ARRAY
!error

edit: copied the wrong testfile first time...this above should be the right one

kgyrtkirk · 2025-04-11T08:44:58Z

furthermore: forgot to mention that the same issue didn't appeared in my IDE - only from a maven run...which is also odd

and even connecting with a remote debugger (-Dmaven.surefire.debug) makes the problem disappear as it just passes

This PR splits up the iq files into smaller files at the testcase level. The files are grouped in subdirectories at the testsuite level. This version also makes the tests run for all 3 engines. The file name of an iq file indicicates what engine(s) the test uses: If all 3 engines expect the same results: *.all.iq: AllDruidEnginesComponentSupplier If one of the enigines expect different results: *.std.iq: StandardComponentSupplier *.dart.iq: DartComponentSupplier *.msq.iq: StandardMSQComponentSupplier

gianm · 2025-04-15T07:48:26Z

I think #17915 might fix the issue you are seeing.

kgyrtkirk · 2025-04-16T06:18:32Z

.../src/test/quidem/org.apache.druid.quidem.QTest/qaUnnest/array_sql_scalar_ipaddr_funcs.all.iq

+                 2)
+       OR s_int IS NULL)
+  AND (ip4_match(c, c) IS NOT NULL);
+No match found


what does No match found mean?
since it supposed to be part of an Exception could we have a more details about the problem?

kgyrtkirk · 2025-04-16T06:21:14Z

...-ut/src/test/quidem/org.apache.druid.quidem.QTest/qaAggFuncs/array_agg_double_test001.all.iq

+!ok
+
+#-------------------------------------------------------------------------
+# TESTCASE: test001 TEST_ID: A02 TYPE: POSITIVE TEST


I think this TYPE: part is redundant: it has no real value as the test has some expectation like !ok - which should document the expected outcome already

kgyrtkirk · 2025-04-16T06:23:48Z

quidem-ut/src/test/quidem/org.apache.druid.quidem.QTest/qaUnnest/array_sql_array_funcs.all.iq

+!ok
+
+#-------------------------------------------------------------------------
+# Total query count 283 Positive tests: 135 Negative tests: 148


can we have more cases and less queries per file?

kgyrtkirk · 2025-04-16T06:25:13Z

.../src/test/quidem/org.apache.druid.quidem.QTest/qaUnnest/array_sql_col_datatype_array.dart.iq

+SELECT c
+FROM test_unnest,
+     unnest(s_int) AS u(c);
+Cannot apply


what does Cannot apply mean here? it should be more readable why this doesn't work

kgyrtkirk · 2025-04-16T06:30:02Z

...t/src/test/quidem/org.apache.druid.quidem.QTest/qaUnnest/array_sql_col_datatype_array.std.iq

+!ok
+
+#-------------------------------------------------------------------------
+# Total query count 43 Positive tests: 37 Negative tests: 6


I don't think these headers are usefull - instead they could cause confusion:
what happens if tomorrow one more NEGATIVE tests will be supported...should these numbers be adjusted as well?

kgyrtkirk · 2025-04-16T06:32:30Z

sql/src/test/quidem/qatests/kttm_nested/data/kttm-nested-v2-2019-08-25.json.gz

large files should not arrive like this! see other comment

kgyrtkirk · 2025-04-16T06:37:15Z

sql/src/test/quidem/qatests/kttm_nested/ingest.json

+        "type" : "local",
+        "baseDir" : "sql/src/test/quidem/qatests/kttm_nested/data",
+        "filter" : "kttm-nested-v2-2019-08-25.json.gz"


we should find a way to access the already accessible kttm-nested-v2-2019-08-25.json on the classpath; and that should be the way to get larger datafiles

What is the equivalent of !use druidtest:///?componentSupplier=KttmNestedComponentSupplier if to use this dataset for AllDruidEnginesComponentSupplier, DartComponentSupplier, StandardMSQComponentSupplier, and StandardComponentSupplier? These tests originally used !use druidtest:///?componentSupplier=KttmNestedComponentSupplier. We need to make them run for all 3 engines, but I couldn't find a way to do so. That's why the dataset was added.

weishiuntsai · 2025-04-22T02:05:06Z

I think #17915 might fix the issue you are seeing.

Tried the latest code. It still OOMed.

[WIP] Tests for debugging Dart and MSQ in Quidem

df97853

Add 2 tests for debugging Dart and Msq in Quidem.

github-actions bot added the Area - Querying label Apr 11, 2025

kgyrtkirk reviewed Apr 16, 2025

View reviewed changes

Merge remote-tracking branch 'apache/master' into debug_quidem_tests

42a47de

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[WIP] Tests for debugging Dart and MSQ in Quidem #17861

[WIP] Tests for debugging Dart and MSQ in Quidem #17861

weishiuntsai commented Apr 2, 2025

kgyrtkirk commented Apr 10, 2025

kgyrtkirk commented Apr 11, 2025 •

edited

Loading

kgyrtkirk commented Apr 11, 2025

gianm commented Apr 15, 2025

kgyrtkirk Apr 16, 2025

kgyrtkirk Apr 16, 2025

kgyrtkirk Apr 16, 2025

kgyrtkirk Apr 16, 2025

kgyrtkirk Apr 16, 2025

kgyrtkirk Apr 16, 2025

kgyrtkirk Apr 16, 2025

weishiuntsai Apr 22, 2025

weishiuntsai commented Apr 22, 2025

[WIP] Tests for debugging Dart and MSQ in Quidem #17861

Are you sure you want to change the base?

[WIP] Tests for debugging Dart and MSQ in Quidem #17861

Conversation

weishiuntsai commented Apr 2, 2025

kgyrtkirk commented Apr 10, 2025

kgyrtkirk commented Apr 11, 2025 • edited Loading

kgyrtkirk commented Apr 11, 2025

gianm commented Apr 15, 2025

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

weishiuntsai commented Apr 22, 2025

kgyrtkirk commented Apr 11, 2025 •

edited

Loading