roachtest: tlp failed

[cockroach-teamcity]

roachtest.tlp failed with artifacts on release-21.2 @ 5afb632f77eee9f09f2adfa2943e1979ec4ebedf:

The test failed on branch=release-21.2, cloud=gce:
test artifacts and logs in: /home/agent/work/.go/src/github.com/cockroachdb/cockroach/artifacts/tlp/run_1
	tlp.go:128,test_runner.go:777: expected unpartitioned count 8 to equal partitioned count 11
		(1) attached stack trace
		  -- stack trace:
		  | github.com/cockroachdb/cockroach/pkg/cmd/roachtest/tests.runTLPQuery.func2
		  | 	/home/agent/work/.go/src/github.com/cockroachdb/cockroach/pkg/cmd/roachtest/tests/tlp.go:191
		  | github.com/cockroachdb/cockroach/pkg/cmd/roachtest/tests.runWithTimeout.func1
		  | 	/home/agent/work/.go/src/github.com/cockroachdb/cockroach/pkg/cmd/roachtest/tests/tlp.go:203
		  | runtime.goexit
		  | 	/usr/local/go/src/runtime/asm_amd64.s:1581
		Wraps: (2) expected unpartitioned count 8 to equal partitioned count 11
		  | sql: SELECT count(*) FROM defaultdb.public.table3 AS tab_18624
		  | SELECT count(*) FROM (SELECT * FROM defaultdb.public.table3 AS tab_18624 WHERE (NOT ('21:17:16.418508'::TIME NOT IN (tab_18624.col3_3,))) UNION ALL SELECT * FROM defaultdb.public.table3 AS tab_18624 WHERE NOT ((NOT ('21:17:16.418508'::TIME NOT IN (tab_18624.col3_3,)))) UNION ALL SELECT * FROM defaultdb.public.table3 AS tab_18624 WHERE ((NOT ('21:17:16.418508'::TIME NOT IN (tab_18624.col3_3,)))) IS NULL)
		Error types: (1) *withstack.withStack (2) *errutil.leafError

Reproduce

See: roachtest README

Same failure on other branches

• roachtest: tlp failed #74675 roachtest: tlp failed [C-test-failure O-roachtest O-robot T-sql-queries branch-master release-blocker]

/cc @cockroachdb/sql-queries

This test on roachdash | Improve this report!

[rytaft]

Not a release blocker, as this problem existed on 21.2.4. The problem seems to be fixed on master.

[cucaroach]

I tried to bisect on master when this was fixed but couldn't get it to fail in the same, every build a couple months old I tried failed like this:

ERROR: value type regnamespace doesn't match type regrole of column "col1_6"
SQLSTATE: 42804
HINT: you will need to rewrite or cast the expression
ERROR: value type regnamespace doesn't match type regrole of column "col1_6"
SQLSTATE: 42804
HINT: you will need to rewrite or cast the expression
Failed running "demo"

@mgartner any sage advice? Should I bail on the bisect idea and just reduce the failure on 21.2?

[mgartner]

I think reducing on 21.2 is a good idea. It should make it easier to identify it as a known bug that we've already fixed, if that is the case.

[cockroach-teamcity]

roachtest.tlp failed with artifacts on release-21.2 @ d7a023225d0db89453ebe3193d59678907e464f3:

The test failed on branch=release-21.2, cloud=gce:
test artifacts and logs in: /home/agent/work/.go/src/github.com/cockroachdb/cockroach/artifacts/tlp/run_1
	tlp.go:128,test_runner.go:777: expected unpartitioned count 3 to equal partitioned count 48
		(1) attached stack trace
		  -- stack trace:
		  | github.com/cockroachdb/cockroach/pkg/cmd/roachtest/tests.runTLPQuery.func2
		  | 	/home/agent/work/.go/src/github.com/cockroachdb/cockroach/pkg/cmd/roachtest/tests/tlp.go:191
		  | github.com/cockroachdb/cockroach/pkg/cmd/roachtest/tests.runWithTimeout.func1
		  | 	/home/agent/work/.go/src/github.com/cockroachdb/cockroach/pkg/cmd/roachtest/tests/tlp.go:203
		  | runtime.goexit
		  | 	/usr/local/go/src/runtime/asm_amd64.s:1581
		Wraps: (2) expected unpartitioned count 3 to equal partitioned count 48
		  | sql: SELECT count(*) FROM defaultdb.public.table1 AS tab_8567
		  | SELECT count(*) FROM (SELECT * FROM defaultdb.public.table1 AS tab_8567 WHERE ('' <= tab_8567.col1_8) UNION ALL SELECT * FROM defaultdb.public.table1 AS tab_8567 WHERE NOT (('' <= tab_8567.col1_8)) UNION ALL SELECT * FROM defaultdb.public.table1 AS tab_8567 WHERE (('' <= tab_8567.col1_8)) IS NULL)
		Error types: (1) *withstack.withStack (2) *errutil.leafError

Reproduce

See: roachtest README

Same failure on other branches

• sql: incorrect splitting lookup span into family spans #76289 roachtest: tlp failed [C-test-failure O-roachtest O-robot T-sql-queries branch-master release-blocker]

/cc @cockroachdb/sql-queries

This test on roachdash | Improve this report!

[yuzefovich]

I'm testing the updated reduce utility.

The first failure is reduced to:

CREATE TABLE table3 (
  col3_0 CHAR,
  col3_1 DECIMAL,
  col3_2 DATE,
  col3_3 TIME,
  col3_4 NAME,
  col3_5 BIT(7),
  col3_6 STRING AS (lower(col3_0)) VIRTUAL,
  UNIQUE (col3_6 ASC),
  INDEX (col3_2 DESC) STORING(col3_5)
);

INSERT
  INTO table3 (col3_1, col3_0, col3_3, col3_4, col3_2)
VALUES (0, ' ':::STRING, '08:29:43.540428', '', '1973-07-30');

UPDATE table3 AS tab_18480 SET col3_0 = 'H', col3_4 = tab_18480.col3_4;

The second failure is reduced to:

CREATE TABLE table1 (
  col1_0 "char",
  col1_1 DATE,
  col1_2 REGTYPE,
  col1_3 DECIMAL,
  col1_4 BIT(8),
  col1_5 BOX2D,
  col1_6 STRING,
  col1_7 STRING AS (CASE WHEN col1_2 IS NULL THEN e'nR\x19P\x0e+\x1a' ELSE '' END) STORED,
  col1_8 STRING AS (lower(col1_0)) VIRTUAL,
  UNIQUE (col1_8 DESC, col1_5 ASC),
  INDEX (col1_1 DESC)
);

INSERT
  INTO table1 (col1_0, col1_1, col1_3, col1_4, col1_5)
VALUES ('', '1986-08-12', 0, B'11001001', 'BOX(-10 -10,10 10)');

INSERT INTO table1 SELECT tab_3020.col1_7, tab_3020.col1_1, 0, 0 FROM table1 AS tab_3020;

DELETE FROM table1;

Both seem like assignment-cast related, so we probably just want to close this issue, but I'll defer to Michael.

[michae2]

Nice! I'll bisect this afternoon and then we can decide what to do.

[michae2]

Yep, both failures were fixed by 17276b5 (assignment casts for INSERTs). Closing.

[mgartner]

Thanks for diagnosing! I’ve just merged #76614 which will prevent TLP from repeatedly finding this bug on 21.2.