[MDEV-10261] Tokudb test failures with --big-test Created: 2016-06-21  Updated: 2016-07-20  Resolved: 2016-07-13

Status: Closed
Project: MariaDB Server
Component/s: Storage Engine - TokuDB, Tests
Affects Version/s: 10.0.25
Fix Version/s: 10.1.16, 10.0.27

Type: Bug Priority: Major
Reporter: Daniel Black Assignee: Sergei Golubchik
Resolution: Fixed Votes: 0
Labels: contribution, foundation, patch

Attachments: File mariadb-10.0-build_and_test.log    

 Description   

--- /source/storage/tokudb/mysql-test/tokudb_parts/r/part_supported_sql_func_tokudb.result	2016-06-20 20:13:56.924740471 +0000
+++ /source/storage/tokudb/mysql-test/tokudb_parts/r/part_supported_sql_func_tokudb.reject	2016-06-20 20:36:19.439745290 +0000
@@ -47,6 +47,7 @@
 -------------------------------------------------------------------------
 ---  Access tables with abs(col1) 
 -------------------------------------------------------------------------
+begin;
 insert into t1 values (5 );
 insert into t1 values (13 );
 insert into t2 values (5 );
@@ -55,6 +56,7 @@
 insert into t3 values (5 );
 insert into t3 values (13 );
 insert into t3 values (17 );
+commit;
 load data infile 'MYSQLTEST_VARDIR/std_data/parts/part_supported_sql_funcs_int_int.inc' into table t4;
 load data infile 'MYSQLTEST_VARDIR/std_data/parts/part_supported_sql_funcs_int_int.inc' into table t5;
 load data infile 'MYSQLTEST_VARDIR/std_data/parts/part_supported_sql_funcs_int_int.inc' into table t6;

Appears to be as a result of adding begin/commit in mysql-test/suite/parts/inc/partition_supported_sql_funcs.inc (https://github.com/MariaDB/server/commit/5ab92b1f#diff-d29bc9df92f5e23ce69ec497dd7da9e0R19).

CURRENT_TEST: tokudb_parts.partition_alter4_tokudb^M
--- /source/storage/tokudb/mysql-test/tokudb_parts/r/partition_alter4_tokudb.result     2016-06-20 20:13:56.948740811 +0000^M
+++ /source/storage/tokudb/mysql-test/tokudb_parts/r/partition_alter4_tokudb.reject     2016-06-20 20:38:13.165343121 +0000^M
@@ -74,7 +74,7 @@^M
   `f_char1` char(20) DEFAULT NULL,^M
   `f_char2` char(20) DEFAULT NULL,^M
   `f_charbig` varchar(1000) DEFAULT NULL^M
-) ENGINE=TokuDB DEFAULT CHARSET=latin1^M
+) ENGINE=TokuDB DEFAULT CHARSET=latin1 `compression`='tokudb_zlib'^M
 /*!50100 PARTITION BY HASH (f_int1)^M
 (PARTITION part_1 ENGINE = TokuDB,^M
  PARTITION part_2 ENGINE = TokuDB) */^M
@@ -533,7 +533,7 @@^M
   `f_char1` char(20) DEFAULT NULL,^M
   `f_char2` char(20) DEFAULT NULL,^M
   `f_charbig` varchar(1000) DEFAULT NULL^M
-) ENGINE=TokuDB DEFAULT CHARSET=latin1^M
+) ENGINE=TokuDB DEFAULT CHARSET=latin1 `compression`='tokudb_zlib'^M
 /*!50100 PARTITION BY KEY (f_int1)^M
 (PARTITION part_1 ENGINE = TokuDB,^M
  PARTITION part_2 ENGINE = TokuDB,^M

Not sure of when this changed

Segfault on shutdown - might be MDEV-9091 will raise separate bug with backtrace



 Comments   
Comment by Daniel Black [ 2016-06-22 ]

In the 10.2 branch these two test branches fail. Looks like the test output just needs re-recording.

12:28:46 CURRENT_TEST: tokudb.rows-32m-rand-insert
12:28:46 --- /source/storage/tokudb/mysql-test/tokudb/r/rows-32m-rand-insert.result	2016-06-22 01:52:18.282578745 +0000
12:28:46 +++ /source/storage/tokudb/mysql-test/tokudb/r/rows-32m-rand-insert.reject	2016-06-22 02:28:45.981158917 +0000
12:28:46 @@ -3,6 +3,7 @@
12:28:46  create table t (id int not null primary key, v longblob not null);
12:28:46  select @@max_allowed_packet into @my_max_allowed_packet;
12:28:46  set global max_allowed_packet=100000000;
12:28:46 +connect conn1,localhost,root,,;
12:28:46  insert ignore into t (id,v) values (floor(rand()*1000),repeat('a',(32*1024*1024-4)*rand()));
12:28:46  insert ignore into t (id,v) values (floor(rand()*1000),repeat('a',(32*1024*1024-4)*rand()));
12:28:46  insert ignore into t (id,v) values (floor(rand()*1000),repeat('a',(32*1024*1024-4)*rand()));
12:28:46 @@ -1003,6 +1004,8 @@
12:28:46  insert ignore into t (id,v) values (floor(rand()*1000),repeat('a',(32*1024*1024-4)*rand()));
12:28:46  insert ignore into t (id,v) values (floor(rand()*1000),repeat('a',(32*1024*1024-4)*rand()));
12:28:46  insert ignore into t (id,v) values (floor(rand()*1000),repeat('a',(32*1024*1024-4)*rand()));
12:28:46 +connection default;
12:28:46 +disconnect conn1;
12:28:46  set global max_allowed_packet=@my_max_allowed_packet;
12:28:46  check table t;
12:28:46  Table	Op	Msg_type	Msg_text
12:28:46 
12:28:46 mysqltest: Result length mismatch
12:28:46 
12:28:46  - saving '/build/build/mysql-test/var/6/log/tokudb.rows-32m-rand-insert/' to '/build/build/mysql-test/var/log/tokudb.rows-32m-rand-insert/'
12:29:16 tokudb.rows-32m-seq-insert               w17 [ fail ]
12:29:16         Test ended at 2016-06-22 02:29:16
12:29:16 
12:29:16 CURRENT_TEST: tokudb.rows-32m-seq-insert
12:29:16 --- /source/storage/tokudb/mysql-test/tokudb/r/rows-32m-seq-insert.result	2016-06-22 01:52:18.282578745 +0000
12:29:16 +++ /source/storage/tokudb/mysql-test/tokudb/r/rows-32m-seq-insert.reject	2016-06-22 02:29:16.229472322 +0000
12:29:16 @@ -3,6 +3,7 @@
12:29:16  create table t (id int not null auto_increment primary key, v longblob not null);
12:29:16  select @@max_allowed_packet into @my_max_allowed_packet;
12:29:16  set global max_allowed_packet=100000000;
12:29:16 +connect conn1,localhost,root,,;
12:29:16  insert into t (v) values (repeat('a',(32*1024*1024-4)*rand()));
12:29:16  insert into t (v) values (repeat('a',(32*1024*1024-4)*rand()));
12:29:16  insert into t (v) values (repeat('a',(32*1024*1024-4)*rand()));
12:29:16 @@ -1003,6 +1004,8 @@
12:29:16  insert into t (v) values (repeat('a',(32*1024*1024-4)*rand()));
12:29:16  insert into t (v) values (repeat('a',(32*1024*1024-4)*rand()));
12:29:16  insert into t (v) values (repeat('a',(32*1024*1024-4)*rand()));
12:29:16 +connection default;
12:29:16 +disconnect conn1;
12:29:16  set global max_allowed_packet=@my_max_allowed_packet;
12:29:16  check table t;
12:29:16  Table	Op	Msg_type	Msg_text
12:29:16 
12:29:16 mysqltest: Result length mismatch

Comment by Elena Stepanova [ 2016-07-01 ]

Apparently it doesn't run in buildbot at all. The only builders where we have default suites to run with --big are fulltests, and they don't build TokuDB because cmake version is too low.

Comment by Rich Prohaska [ 2016-07-02 ]

See https://github.com/MariaDB/server/pull/199.

Comment by Sergey Vojtovich [ 2016-07-05 ]

Removed 10.0 from fix version since PR is against 10.1, so that this task fits into current 10.1 sprint. Still it should be fixed in 10.0 I believe.

Comment by Daniel Black [ 2016-07-10 ]

so fixed in https://github.com/MariaDB/server/pull/199#issuecomment-231361145

hoping you can update cmake on a buildbot to make these tests run.

Generated at Thu Feb 08 07:40:52 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.