[MDEV-11614] Syslog messages: "InnoDB: Log sequence number at the start 759654123 and the end 0 do not match." Created: 2016-12-20  Updated: 2017-01-22  Resolved: 2017-01-22

Status: Closed
Project: MariaDB Server
Component/s: Storage Engine - InnoDB, Storage Engine - XtraDB
Affects Version/s: 10.1.19
Fix Version/s: 10.1.22

Type: Bug Priority: Major
Reporter: Sergey Antonyuk Assignee: Jan Lindström (Inactive)
Resolution: Fixed Votes: 0
Labels: None
Environment:

Debian Linux 3.10.0 x86_64



 Description   

The following messages appears in syslog about once in an hour:

Dec 20 15:54:33 www mysqld: 2016-12-20 15:54:33 140047920658176 [Note] InnoDB: Log sequence number at the start 759654123 and the end 0 do not match.
Dec 20 15:54:33 www mysqld: 2016-12-20 15:54:33 140047920658176 [Note] InnoDB: Log sequence number at the start 759654123 and the end 0 do not match.
Dec 20 15:54:33 www mysqld: 2016-12-20 15:54:33 140047920658176 [Note] InnoDB: Log sequence number at the start 2785176653 and the end 0 do not match.
Dec 20 15:54:33 www mysqld: 2016-12-20 15:54:33 140047920658176 [Note] InnoDB: Log sequence number at the start 735407746 and the end 0 do not match.
Dec 20 15:54:33 www mysqld: 2016-12-20 15:54:33 140047920658176 [Note] InnoDB: Log sequence number at the start 1336596246 and the end 0 do not match.
Dec 20 15:54:33 www mysqld: 2016-12-20 15:54:33 140047920658176 [Note] InnoDB: Log sequence number at the start 3570592028 and the end 0 do not match.

{/quote}

Scenario is unknown.

Please let me know whether it's a bug or configuration issue.



 Comments   
Comment by Sergey Antonyuk [ 2016-12-20 ]

It looks like the error appears on simple SELECT all records request from InnoDB `page_compressed`=1 table with the following structure:

CREATE TABLE `o` (
`id` int(10) unsigned NOT NULL AUTO_INCREMENT,
`ts` timestamp NOT NULL DEFAULT CURRENT_TIMESTAMP,
`up` int(10) unsigned NOT NULL,
`ud` int(10) unsigned NOT NULL,
`un` varchar(40) NOT NULL,
`dr` text NOT NULL,
`de` text NOT NULL,
`t` enum('ss','rs','da','ws','sg','sp','rf','os','ls','la','lt','nt','ts','lao') NOT NULL,
`uss` enum('ne','pe','i','d') NOT NULL DEFAULT 'ne',
`ui` text,
`uid` int(10) unsigned DEFAULT NULL,
`ao` text,
`eo` mediumblob,
PRIMARY KEY (`id`)
) ENGINE=InnoDB AUTO_INCREMENT=36830 DEFAULT CHARSET=utf8 `page_compressed`=1

This table contains 37k of records. It has been created earlier in MySQL 5.1, upgraded to MariaDB and altered to page compressed version. Current data length = 23Mb, index length 0, data free 4Mb, avg row length 704, row format dynamic, version 10.

Comment by Vladislav Vaintroub [ 2017-01-20 ]

I have the same effect using page_compression, zlib algorithm with largeish data set (DBT3, factor1)
that floods the log quite a lot.

Comment by Jan Lindström (Inactive) [ 2017-01-22 ]

commit b7b4c332c0869676ef69276ddaef4ac4521cb04f
Author: Jan Lindström <jan.lindstrom@mariadb.com>
Date: Sun Jan 22 08:44:04 2017 +0200

MDEV-11614: Syslog messages: "InnoDB: Log sequence number

at the start 759654123 and the end 0 do not match."

For page compressed and encrypted tables log sequence
number at end is not stored, thus disable this message
for them.

Generated at Thu Feb 08 07:51:19 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.