4.6.1. innochecksum — Offline InnoDB File Checksum Utility

innochecksum prints checksums for InnoDB files. This tool reads an InnoDB tablespace file, calculates the checksum for each page, compares the calculated checksum to the stored checksum, and reports mismatches, which indicate damaged pages. It was originally developed to speed up verifying the integrity of tablespace files after power outages but can also be used after file copies. Because checksum mismatches will cause InnoDB to deliberately shut down a running server, it can be preferable to use this tool rather than waiting for a server in production usage to encounter the damaged pages.

innochecksum cannot be used on tablespace files that the server already has open. For such files, you should use CHECK TABLE to check tables within the tablespace. Attempting to run innochecksum on a tablespace that the server already has open will result in an "Unable to lock file" error.

If checksum mismatches are found, you would normally restore the tablespace from backup or start the server and attempt to use mysqldump to make a backup of the tables within the tablespace.

Invoke innochecksum like this:

shell> innochecksum [options]
        file_name

innochecksum supports the following options. For options that refer to page numbers, the numbers are zero-based.

--help, -?

Displays command line help. Example usage:
```
shell> innochecksum --help
```
--info, -I

Synonym for --help. Displays command line help. Example usage:
```
shell> innochecksum --info
```
--version, -V

Displays version information. Example usage:
```
shell> innochecksum --version
```

--verbose, -v

Verbose mode; print a progress indicator every five seconds. Example usage:

shell> innochecksum --verbose ../data/test/tab1.ibd

Sample output:

Variables (--variable-name=value)and boolean options {FALSE|TRUE}  Value (after reading options)--------------------------------- -----------------------------verbose                           TRUEdebug                             FALSEcount                             FALSEstart-page                        0end-page                          0page                              0strict-check                      crc32no-check                          FALSEallow-mismatches                  0write                             crc32page-type-summary                 FALSEpage-type-dump                    (No default value)log                               (No default value)

--count, -c

Print a count of the number of pages in the file and exit. Example usage:
```
shell> innochecksum --count ../data/test/tab1.ibd
```

--start-page=num, -s num

Start at this page number. Example usage:

shell> innochecksum --start-page=600 ../data/test/tab1.ibd

or:

shell> innochecksum -s 600 ../data/test/tab1.ibd

--end-page=num, -e num

End at this page number. Example usage:

shell> innochecksum --end-page=700 ../data/test/tab1.ibd

or:

shell> innochecksum --p 700 ../data/test/tab1.ibd

--page=num, -p num

Check only this page number. Example usage:
```
shell> innochecksum --page=701 ../data/test/tab1.ibd
```
--strict-check, -C

Specify a strict checksum algorithm. Options include innodb, crc32, and none.

In this example, the innodb checksum algorithm is specified:
```
shell> innochecksum --strict-check=innodb ../data/test/tab1.ibd
```
In this example, the crc32 checksum algorithm is specified:
```
shell> innochecksum -C crc32 ../data/test/tab1.ibd
```
The following conditions apply:
- If you do not specify the --strict-check option, innochecksum validates against innodb, crc32 and none.
- If you specify the none option, only checksums generated by none are allowed.
- If you specify the innodb option, only checksums generated by innodb are allowed.
- If you specify the crc32 option, only checksums generated by crc32 are allowed.
--no-check, -n

Ignore the checksum verification when rewriting a checksum. This option may only be used with the innochecksum --write option. If the --write option is not specified, innochecksum will terminate.

In this example, an innodb checksum is rewritten to replace an invalid checksum:
```
shell> innochecksum --no-check --write innodb ../data/test/tab1.ibd
```
--allow-mismatches, -a

The maximum number of checksum mismatches allowed before innochecksum terminates. The default setting is 0. If --allow-mismatches=N, where N>=0, N mismatches are permitted and innochecksum terminates at N+1. When --allow-mismatches is set to 0, innochecksum terminates on the first checksum mismatch.

In this example, an existing innodb checksum is rewritten to set --allow-mismatches to 1.
```
shell> innochecksum --allow-mismatches=1 --write innodb ../data/test/tab1.ibd
```
With --allow-mismatches set to 1, if there is a mismatch at page 600 and another at page 700 on a file with 1000 pages, the checksum is updated for pages 0-599 and 601-699. Because --allow-mismatches is set to 1, the checksum tolerates the first mismatch and terminates on the second mismatch, leaving page 600 and pages 700-999 unchanged.
--write=name, -w num

Rewrite a checksum. When rewriting an invalid checksum, the --no-check option must be used together with the --write option. The --no-check option tells innochecksum to ignore verification of the invalid checksum. You do not have to specify the --no-check option if the current checksum is valid.

An algorithm must be specified when using the --write option. Possible values for the --write option are:
- innodb: A checksum calculated in software, using the original algorithm from InnoDB.
- crc32: A checksum calculated using the crc32 algorithm, possibly done with a hardware assist.
- none: A constant number.
The --write option rewrites entire pages to disk. If the new checksum is identical to the existing checksum, the new checksum is not written to disk in order to minimize I/O.

innochecksum obtains an exclusive lock when the --write option is used.

In this example, a crc32 checksum is written for tab1.ibd:
```
shell> innochecksum -w crc32 ../data/test/tab1.ibd
```
In this example, a crc32 checksum is rewritten to replace an invalid crc32 checksum:
```
shell> innochecksum --no-check --write crc32 ../data/test/tab1.ibd
```

--page-type-summary, -S

Display a count of each page type in a tablespace. Example usage:

shell> innochecksum --page-type-summary ../data/test/tab1.ibd

Sample output for --page-type-summary:

File::../data/test/tab1.ibd================PAGE TYPE SUMMARY==============#PAGE_COUNT	PAGE_TYPE===============================================       2	Index page       0	Undo log page       1	Inode page       0	Insert buffer free list page       2	Freshly allocated page       1	Insert buffer bitmap       0	System page       0	Transaction system page       1	File Space Header       0	Extent descriptor page       0	BLOB page       0	Compressed BLOB page       0	Other type of page===============================================Additional information:Undo page type: 0 insert, 0 update, 0 otherUndo page state: 0 active, 0 cached, 0 to_free, 0 to_purge, 0 prepared, 0 other

--page-type-dump, -D

Dump the page type information for each page in a tablespace to stderr or stdout. Example usage:
```
shell> innochecksum --page-type-dump /tmp/a.txt ../data/test/tab1.ibd
```
--log, -l

Log output for the innochecksum tool. A log file name must be provided. Log output contains checksum values for each tablespace page. For uncompressed tables, LSN values are also provided. The --log replaces the --debug option, which was available in earlier releases. Example usage:
```
shell> innochecksum --log=/tmp/log.txt ../data/test/tab1.ibd
```
or:
```
shell> innochecksum -l /tmp/log.txt ../data/test/tab1.ibd
```
"-" option.

Specify the "-" option to read from standard input. If the "-" option is missing when "read from standard in" is expected, innochecksum reports an error.
```
shell> cat t1.ibd | ./bin/innochecksum  -
```