Skip to content

Commit 8589b8a

Browse files
craig[bot]stevendanna
andcommitted
Merge #146792
146792: logictest: update buffered writes TODO in txn_retry test r=yuzefovich a=stevendanna Here we update the error references for txn_retry now that we understand this test, closing the specific bug for this test in favor of the more general bug. Informs #146732 Fixes #144278 Release note: None Co-authored-by: Steven Danna <[email protected]>
2 parents d59ff4e + 232bee2 commit 8589b8a

File tree

1 file changed

+13
-4
lines changed
  • pkg/sql/logictest/testdata/logic_test

1 file changed

+13
-4
lines changed

pkg/sql/logictest/testdata/logic_test/txn_retry

Lines changed: 13 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -1,11 +1,20 @@
1-
# TODO(#144278): remove this or adjust the test.
2-
statement ok
3-
SET kv_transaction_buffered_writes_enabled = false;
4-
51
# Check that we auto-retry pushed transactions which can't be refreshed - if
62
# they're pushed while we can still auto-retry them.
73
subtest autoretry-on-push-first-batch
84

5+
# TODO(#146732): This test is testing a specific retry path that is
6+
# only used when a statement has resulted in a WriteTimestamp push,
7+
# can't be refreshed by the span refresher interceptor, and also
8+
# hasn't produced an error. In buffered writes, we don't detect this
9+
# WriteTimestamp push until the COMMIT and are no longer eligible for
10+
# the retry path tested here.
11+
#
12+
# It may seem like we could change the read-write conflict to a
13+
# write-write conflict. But the write-write conflict generates an
14+
# error which means it wouldn't test the same retry path.
15+
statement ok
16+
SET kv_transaction_buffered_writes_enabled = false;
17+
918
statement ok
1019
CREATE TABLE test_retry (
1120
k INT PRIMARY KEY

0 commit comments

Comments
 (0)