[MDEV-27160] main.long_unique failed on ppc64el on Ubuntu 21.10 autopkgtest Created: 2021-12-03  Updated: 2021-12-03  Resolved: 2021-12-03

Status: Closed
Project: MariaDB Server
Component/s: Tests
Affects Version/s: None
Fix Version/s: 10.4.23, 10.5.14, 10.6.6, 10.7.2

Type: Bug Priority: Major
Reporter: Otto Kekäläinen Assignee: Marko Mäkelä
Resolution: Fixed Votes: 0
Labels: None

Issue Links:
Relates
relates to MDEV-371 Unique indexes for blobs Closed

 Description   

While uploading Mariadb 10.5.13 to Ubuntu 21.10 Impish the following failure in autopkgtests (that runs the mtr among others) was noticed on platform ppc64el:

main.long_unique                         w1 [ fail ]
        Test ended at 2021-12-02 15:01:58
 
CURRENT_TEST: main.long_unique
mysqltest: At line 409: query 'insert into t1 values(concat(repeat('sachin',10000000),'1'),concat(repeat('sachin',10000000),'1'),
concat(repeat('sachin',10000000),'1'))' failed: 2013: Lost connection to MySQL server during query
 
The result from queries just before the failure was:
< snip >
Field	Type	Null	Key	Default	Extra
a	longblob	YES	UNI	NULL	
b	longblob	YES	MUL	NULL	
c	longblob	YES		NULL	
show create table t1;
Table	Create Table
t1	CREATE TABLE `t1` (
  `a` longblob DEFAULT NULL,
  `b` longblob DEFAULT NULL,
  `c` longblob DEFAULT NULL,
  UNIQUE KEY `a` (`a`) USING HASH,
  UNIQUE KEY `b` (`b`,`c`) USING HASH
) ENGINE=MyISAM DEFAULT CHARSET=latin1
show keys from t1;
Table	Non_unique	Key_name	Seq_in_index	Column_name	Collation	Cardinality	Sub_part	Packed	Null	Index_type	Comment	Index_comment
t1	0	a	1	a	A	NULL	NULL	NULL	YES	HASH		
t1	0	b	1	b	A	NULL	NULL	NULL	YES	HASH		
t1	0	b	2	c	A	NULL	NULL	NULL	YES	HASH		
insert into t1 values(concat(repeat('sachin',10000000),'1'),concat(repeat('sachin',10000000),'1'),
concat(repeat('sachin',10000000),'1'));
 
More results from queries before failure can be found in /tmp/tmp.P0sky4MUVI/var/1/log/long_unique.log
 
 
Server [mysqld.1 - pid: 4618, winpid: 4618, exit: 256] failed during test run
Server log from this test:
----------SERVER LOG START-----------
----------SERVER LOG END-------------

For more details please see https://bugs.launchpad.net/ubuntu/+source/mariadb-10.5/+bug/1951709/comments/5



 Comments   
Comment by Marko Mäkelä [ 2021-12-03 ]

Could it be simple exceeding the available memory? We got the following happening on AIX:

10.5 78bd7d86a4a0b3733ba38373213645a6b3a9b9c6

main.long_unique                         w5 [ fail ]
        Test ended at 2021-06-18 00:14:45
 
CURRENT_TEST: main.long_unique
mysqltest: At line 409: query 'insert into t1 values(concat(repeat('sachin',10000000),'1'),concat(repeat('sachin',10000000),'1'),
concat(repeat('sachin',10000000),'1'))' failed: 1041: Out of memory.

It feels steep to construct a row of two 60,000,001-byte columns, but the purpose of this test section seems to be to test long BLOB values. The storage engine likely is Aria, because one is not being explicitly specified in the test.

The actual memory consumption could be several times the 120MB. And this section of the test will attempt to insert 4 such rows.
Perhaps this part of the test should be moved to a separate test file that would only be run with --big-test.

Comment by Marko Mäkelä [ 2021-12-03 ]

I moved the memory-hogging part of the test to a new test main.long_unique_big.

Generated at Thu Feb 08 09:50:48 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.