Repairing a crashed MySQL Table Revision as of Sunday, 20 December 2015 at 19:56 UTC

You may be chugging along, doing your thing, logging on to MySQL, and
issuing something like this:

 mysql> select * from testdatabase.user;

Your smile may lose its luster when you’re confronted with this:

   ERROR 145 (HY000): Table './testdatabase/user' is marked as crashed and should be repaired

Don’t worry. There are a few things you can do to fix this.

When your job is easy

Usually, issuing this should fix the issue:

 mysql> repair table testdatabase.user;

This, if all goes well, will produce output sort of like this:

 +----------------------+--------+----------+------------------------------------------+
 | Table                | Op     | Msg_type | Msg_text                                 |
 +----------------------+--------+----------+------------------------------------------+
 | testdatabase.user    | repair | warning  | Number of rows changed from 0 to 1151697 |
 | testdatabase.user    | repair | status   | OK                                       |
 +----------------------+--------+----------+------------------------------------------+
 2 rows in set (2 min 7.91 sec)

When your job gets harder

It gets frustrating when you get something like this:

 +------------+--------+----------+------------------------------------------------------------------+
 | Table      | Op     | Msg_type | Msg_text                                                         |
 +------------+--------+----------+------------------------------------------------------------------+
 | mysql.user | repair | Error    | Table './testdatabase/user' is marked as crashed and should be repaired | 
 | mysql.user | repair | Error    | Table 'user' is marked as crashed and should be repaired         | 
 | mysql.user | repair | status   | Table is already up to date                                      | 
 +------------+--------+----------+------------------------------------------------------------------+
 3 rows in set (0.00 sec)

DON’T PANIC. This can be fixed by quitting MySQL, and doing this:

 cd /var/lib/mysql/testdatabase
 myisamchk --safe-recover user

“user”, of course, is the table you wanted to recover. If the Gods have
mercy, you will see something like this:

 Warning: option 'key_buffer_size': unsigned value 18446744073709551615 adjusted to 4294963200
 Warning: option 'read_buffer_size': unsigned value 18446744073709551615 adjusted to 4294967295
 Warning: option 'write_buffer_size': unsigned value 18446744073709551615 adjusted to 4294967295
 Warning: option 'sort_buffer_size': unsigned value 18446744073709551615 adjusted to 4294967295
 - recovering (with keycache) MyISAM-table 'user'
 Data records: 81
 Duplicate key  1 for record at       7918 against new record at       3959
 Duplicate key  1 for record at       8132 against new record at       3638
 Duplicate key  1 for record at       8239 against new record at       1498
 Duplicate key  1 for record at       8560 against new record at       4280
 Data records: 77
 myisamchk: warning: 4 records have been removed

Ta da! Assuming, of course, that your table’s the default MyISAM type.

Category:Nikhil’s Notes
Category:From a past sysadmin
life