Elena Stepanova added a comment - 2017-08-05 22:35 - edited The failure starts from the timed out sync point. We don't see it normally because warnings and result sets are disabled there: connection con2; Table Op Msg_type Msg_text dbt3_s001.lineitem analyze status Engine-independent statistics collected +dbt3_s001.lineitem analyze Warning debug sync point wait timed out dbt3_s001.lineitem analyze status Table is already up to date connection default; disconnect con1; @@ -182,7 +183,6 @@ dbt3_s001 lineitem i_l_orderkey_quantity 2 1.0404 dbt3_s001 lineitem i_l_partkey 1 30.0250 dbt3_s001 lineitem i_l_receiptdate 1 2.6477 -dbt3_s001 lineitem i_l_shipdate 1 2.6500 dbt3_s001 lineitem i_l_suppkey 1 600.5000 dbt3_s001 lineitem i_l_suppkey_partkey 1 30.0250 dbt3_s001 lineitem i_l_suppkey_partkey 2 8.5786 @@ -197,7 +197,6 @@ dbt3_s001 lineitem i_l_orderkey_quantity 2 1.0404 dbt3_s001 lineitem i_l_partkey 1 30.0250 dbt3_s001 lineitem i_l_receiptdate 1 2.6477 -dbt3_s001 lineitem i_l_shipdate 1 2.6500 dbt3_s001 lineitem i_l_suppkey 1 600.5000 dbt3_s001 lineitem i_l_suppkey_partkey 1 30.0250 dbt3_s001 lineitem i_l_suppkey_partkey 2 8.5786 +-----+------+-----------------+-----------+---------+------+------------------------------------------------+---------------------------------------------------------------------------+----------+ | Id | User | Host | db | Command | Time | State | Info | Progress | +-----+------+-----------------+-----------+---------+------+------------------------------------------------+---------------------------------------------------------------------------+----------+ | 316 | root | localhost | dbt3_s001 | Sleep | 69 | | NULL | 0.000 | | 319 | root | localhost | dbt3_s001 | Sleep | 69 | | NULL | 0.000 | | 320 | root | localhost | dbt3_s001 | Query | 69 | debug sync point: statistics_collection_start2 | analyze table lineitem persistent for columns() indexes (i_l_receiptdate) | 0.000 | | 322 | root | localhost:40710 | NULL | Query | 0 | init | show processlist | 0.000 | +-----+------+-----------------+-----------+---------+------+------------------------------------------------+---------------------------------------------------------------------------+----------+ However, sync points are irrelevant. The attached simplified version of the same test without sync points still hits the same error. CURRENT_TEST: bug.mdev13266 --- /data/bld/10.2/mysql-test/suite/bug/mdev13266.result 2017-08-06 19:15:00.826455463 +0300 +++ /data/bld/10.2/mysql-test/suite/bug/mdev13266.reject 2017-08-06 19:52:47.598437878 +0300 @@ -32,5 +32,4 @@ select * from mysql.index_stats where table_name='lineitem' order by index_name, prefix_arity; db_name table_name index_name prefix_arity avg_frequency dbt3_s001 lineitem i_l_partkey 1 30.0250 -dbt3_s001 lineitem i_l_shipdate 1 2.6500 DROP DATABASE dbt3_s001;   mysqltest: Result length mismatch I think it's not a test problem, but a genuine race condition in the code, which requires proper debugging.

Alice Sherepa added a comment - 2017-10-28 16:42 now on 10.3 http://buildbot.askmonty.org/buildbot/builders/kvm-fulltest2/builds/10176/steps/test_2/logs/stdio

Aleksey Midenkov added a comment - 2019-10-17 08:07 Reproduced locally on 10.5.

Varun Gupta (Inactive) added a comment - 2020-04-12 15:39 Patch http://lists.askmonty.org/pipermail/commits/2020-April/014242.html

Sergei Petrunia added a comment - 2020-04-30 20:46 Ok to push after the request in the review email is done.