Skip to content



pt-table-checksum - Verify MySQL replication integrity.



pt-table-checksum [OPTIONS] [DSN]

pt-table-checksum performs an online replication consistency check by executing checksum queries on the master, which produces different results on replicas that are inconsistent with the master. The optional DSN specifies the master host. The tool’s “EXIT STATUS” is non-zero if any differences are found, or if any warnings or errors occur.

The following command will connect to the replication master on localhost, checksum every table, and report the results on every detected replica:


This tool is focused on finding data differences efficiently. If any data is different, you can resolve the problem with pt-table-sync.


Percona Toolkit is mature, proven in the real world, and well tested, but all database tools can pose a risk to the system and the database server. Before using this tool, please:

  • Read the tool’s documentation

  • Review the tool’s known “BUGS”

  • Test the tool on a non-production server

  • Backup your production server and verify the backups



pt-table-checksum is designed to do the right thing by default in almost every case. When in doubt, use --explain to see how the tool will checksum a table. The following is a high-level overview of how the tool functions.

In contrast to older versions of pt-table-checksum, this tool is focused on a single purpose, and does not have a lot of complexity or support many different checksumming techniques. It executes checksum queries on only one server, and these flow through replication to re-execute on replicas. If you need the older behavior, you can use Percona Toolkit version 1.0.

pt-table-checksum connects to the server you specify, and finds databases and tables that match the filters you specify (if any). It works one table at a time, so it does not accumulate large amounts of memory or do a lot of work before beginning to checksum. This makes it usable on very large servers. We have used it on servers with hundreds of thousands of databases and tables, and trillions of rows. No matter how large the server is, pt-table-checksum works equally well.

One reason it can work on very large tables is that it divides each table into chunks of rows, and checksums each chunk with a single REPLACE..SELECT query. It varies the chunk size to make the checksum queries run in the desired amount of time. The goal of chunking the tables, instead of doing each table with a single big query, is to ensure that checksums are unintrusive and don’t cause too much replication lag or load on the server. That’s why the target time for each chunk is 0.5 seconds by default.

The tool keeps track of how quickly the server is able to execute the queries, and adjusts the chunks as it learns more about the server’s performance. It uses an exponentially decaying weighted average to keep the chunk size stable, yet remain responsive if the server’s performance changes during checksumming for any reason. This means that the tool will quickly throttle itself if your server becomes heavily loaded during a traffic spike or a background task, for example.

Chunking is accomplished by a technique that we used to call “nibbling” in other tools in Percona Toolkit. It is the same technique used for pt-archiver, for example. The legacy chunking algorithms used in older versions of pt-table-checksum are removed, because they did not result in predictably sized chunks, and didn’t work well on many tables. All that is required to divide a table into chunks is an index of some sort (preferably a primary key or unique index). If there is no index, and the table contains a suitably small number of rows, the tool will checksum the table in a single chunk.

pt-table-checksum has many other safeguards to ensure that it does not interfere with any server’s operation, including replicas. To accomplish this, pt-table-checksum detects replicas and connects to them automatically. (If this fails, you can give it a hint with the --recursion-method option.)

The tool monitors replicas continually. If any replica falls too far behind in replication, pt-table-checksum pauses to allow it to catch up. If any replica has an error, or replication stops, pt-table-checksum pauses and waits. In addition, pt-table-checksum looks for common causes of problems, such as replication filters, and refuses to operate unless you force it to. Replication filters are dangerous, because the queries that pt-table-checksum executes could potentially conflict with them and cause replication to fail.

pt-table-checksum verifies that chunks are not too large to checksum safely. It performs an EXPLAIN query on each chunk, and skips chunks that might be larger than the desired number of rows. You can configure the sensitivity of this safeguard with the --chunk-size-limit option. If a table will be checksummed in a single chunk because it has a small number of rows, then pt-table-checksum additionally verifies that the table isn’t oversized on replicas. This avoids the following scenario: a table is empty on the master but is very large on a replica, and is checksummed in a single large query, which causes a very long delay in replication.

There are several other safeguards. For example, pt-table-checksum sets its session-level innodb_lock_wait_timeout to 1 second, so that if there is a lock wait, it will be the victim instead of causing other queries to time out. Another safeguard checks the load on the database server, and pauses if the load is too high. There is no single right answer for how to do this, but by default pt-table-checksum will pause if there are more than 25 concurrently executing queries. You should probably set a sane value for your server with the --max-load option.

Checksumming usually is a low-priority task that should yield to other work on the server. However, a tool that must be restarted constantly is difficult to use. Thus, pt-table-checksum is very resilient to errors. For example, if the database administrator needs to kill pt-table-checksum’s queries for any reason, that is not a fatal error. Users often run pt-kill to kill any long-running checksum queries. The tool will retry a killed query once, and if it fails again, it will move on to the next chunk of that table. The same behavior applies if there is a lock wait timeout. The tool will print a warning if such an error happens, but only once per table. If the connection to any server fails, pt-table-checksum will attempt to reconnect and continue working.

If pt-table-checksum encounters a condition that causes it to stop completely, it is easy to resume it with the --resume option. It will begin from the last chunk of the last table that it processed. You can also safely stop the tool with CTRL-C. It will finish the chunk it is currently processing, and then exit. You can resume it as usual afterwards.

After pt-table-checksum finishes checksumming all of the chunks in a table, it pauses and waits for all detected replicas to finish executing the checksum queries. Once that is finished, it checks all of the replicas to see if they have the same data as the master, and then prints a line of output with the results. You can see a sample of its output later in this documentation.

The tool prints progress indicators during time-consuming operations. It prints a progress indicator as each table is checksummed. The progress is computed by the estimated number of rows in the table. It will also print a progress report when it pauses to wait for replication to catch up, and when it is waiting to check replicas for differences from the master. You can make the output less verbose with the --quiet option.

If you wish, you can query the checksum tables manually to get a report of which tables and chunks have differences from the master. The following query will report every database and table with differences, along with a summary of the number of chunks and rows possibly affected:

SELECT db, tbl, SUM(this_cnt) AS total_rows, COUNT(*) AS chunks
FROM percona.checksums
 master_cnt <> this_cnt
 OR master_crc <> this_crc
 OR ISNULL(master_crc) <> ISNULL(this_crc))
GROUP BY db, tbl;

The table referenced in that query is the checksum table, where the checksums are stored. Each row in the table contains the checksum of one chunk of data from some table in the server.

Version 2.0 of pt-table-checksum is not backwards compatible with pt-table-sync version 1.0. In some cases this is not a serious problem. Adding a “boundaries” column to the table, and then updating it with a manually generated WHERE clause, may suffice to let pt-table-sync version 1.0 interoperate with pt-table-checksum version 2.0. Assuming an integer primary key named ‘id’, You can try something like the following:

ALTER TABLE checksums ADD boundaries VARCHAR(500);
UPDATE checksums
 SET boundaries = COALESCE(CONCAT('id BETWEEN ', lower_boundary,
    ' AND ', upper_boundary), '1=1');

Take into consideration that by default, pt-table-checksum use CRC32 checksums. CRC32 is not a cryptographic algorithm and for that reason it is prone to have collisions. On the other hand, CRC32 algorithm is faster and less CPU-intensive than MD5 and SHA1.

Related reading material: Percona Toolkit UDFs: How to avoid hash collisions when using MySQL’s CRC32 function:


Replicas using row-based replication

pt-table-checksum requires statement-based replication, and it sets binlog_format=STATEMENT on the master, but due to a MySQL limitation replicas do not honor this change. Therefore, checksums will not replicate past any replicas using row-based replication that are masters for further replicas.

The tool automatically checks the binlog_format on all servers. See --[no]check-binlog-format .

(Bug 899415)

Schema and table differences

The tool presumes that schemas and tables are identical on the master and all replicas. Replication will break if, for example, a replica does not have a schema that exists on the master (and that schema is checksummed), or if the structure of a table on a replica is different than on the master.

Percona XtraDB Cluster

pt-table-checksum works with Percona XtraDB Cluster (PXC) 5.5.28-23.7 and newer. The number of possible Percona XtraDB Cluster setups is large given that it can be used with regular replication as well. Therefore, only the setups listed below are supported and known to work. Other setups, like cluster to cluster, are not support and probably don’t work.

Except where noted, all of the following supported setups require that you use the dsn method for --recursion-method to specify cluster nodes. Also, the lag check (see “REPLICA CHECKS”) is not performed for cluster nodes.

Single cluster

The simplest PXC setup is a single cluster: all servers are cluster nodes, and there are no regular replicas. If all nodes are specified in the DSN table (see --recursion-method), then you can run the tool on any node and any diffs on any other nodes will be detected.

All nodes must be in the same cluster (have the same wsrep_cluster_name value), else the tool exits with an error. Although it’s possible to have different clusters with the same name, this should not be done and is not supported. This applies to all supported setups.

Single cluster with replicas

Cluster nodes can also be regular masters and replicate to regular replicas. However, the tool can only detect diffs on a replica if ran on the replica’s “master node”. For example, if the cluster setup is,

node1 <-> node2 <-> node3
            |         |
            |         +-> replica3
            +-> replica2

you can detect diffs on replica3 by running the tool on node3, but to detect diffs on replica2 you must run the tool again on node2. If you run the tool on node1, it will not detect diffs on either replica.

Currently, the tool does not detect this setup or warn about replicas that cannot be checked (e.g. replica2 when running on node3).

Replicas in this setup are still subject to --[no]check-binlog-format.

Master to single cluster

It is possible for a regular master to replicate to a cluster, as if the cluster were one logical slave, like:

master -> node1 <-> node2 <-> node3

The tool supports this setup but only if ran on the master and if all nodes in the cluster are consistent with the “direct replica” (node1 in this example) of the master. For example, if all nodes have value “foo” for row 1 but the master has value “bar” for the same row, this diff will be detected. Or if only node1 has this diff, it will also be detected. But if only node2 or node3 has this diff, it will not be detected. Therefore, this setup is used to check that the master and the cluster as a whole are consistent.

In this setup, the tool can automatically detect the “direct replica” (node1) when ran on the master, so you do not have to use the dsn method for --recursion-method because node1 will represent the entire cluster, which is why all other nodes must be consistent with it.

The tool warns when it detects this setup to remind you that it only works when used as described above. These warnings do not affect the exit status of the tool; they’re only reminders to help avoid false-positive results.

RocksDB support

Due to the limitations in the RocksDB engine like not supporting binlog_format=STATEMENT or they way RocksDB handles Gap locks, pt-table-checksum will skip tables using RocksDB engine. More Information: (


The tool prints tabular results, one line per table:

10-20T08:36:50      0      0   200      0       1       0   0.005 db1.tbl1
10-20T08:36:50      0      0   603      3       7       0   0.035 db1.tbl2
10-20T08:36:50      0      0    16      0       1       0   0.003 db2.tbl3
10-20T08:36:50      0      0   600      0       6       0   0.024 db2.tbl4

Errors, warnings, and progress reports are printed to standard error. See also --quiet.

Each table’s results are printed when the tool finishes checksumming the table. The columns are as follows:


The timestamp (without the year) when the tool finished checksumming the table.


The number of errors and warnings that occurred while checksumming the table. Errors and warnings are printed to standard error while the table is in progress.


The number of chunks that differ from the master on one or more replicas. If --no-replicate-check is specified, this column will always have zeros. If --replicate-check-only is specified, then only tables with differences are printed.


The number of rows selected and checksummed from the table. It might be different from the number of rows in the table if you use the –where option.


The maximum number of differences per chunk. If a chunk has 2 different rows and another chunk has 3 different rows, this value will be 3.


The number of chunks into which the table was divided.


The number of chunks that were skipped due one or more of these problems:

* MySQL not using the --chunk-index
* MySQL not using the full chunk index (--[no]check-plan)
* Chunk size is greater than --chunk-size * --chunk-size-limit
* Lock wait timeout exceeded (--retries)
* Checksum query killed (--retries)

As of pt-table-checksum 2.2.5, skipped chunks cause a non-zero “EXIT STATUS”.


The time elapsed while checksumming the table.


The database and table that was checksummed.

If --replicate-check-only is specified, only checksum differences on detected replicas are printed. The output is different: one paragraph per replica, one checksum difference per line, and values are separated by spaces:

Differences on h=,P=12346
db1.tbl1 1 0 1 PRIMARY 1 100
db1.tbl1 6 0 1 PRIMARY 501 600

Differences on h=,P=12347
db1.tbl1 1 0 1 PRIMARY 1 100
db2.tbl2 9 5 0 PRIMARY 101 200

The first line of a paragraph indicates the replica with differences. In this example there are two: h=,P=12346 and h=,P=12347. The columns are as follows:


The database and table that differs from the master.


The chunk number of the table that differs from the master.


The number of chunk rows on the replica minus the number of chunk rows on the master.


1 if the CRC of the chunk on the replica is different than the CRC of the chunk on the master, else 0.


The index used to chunk the table.


The index values that define the lower boundary of the chunk.


The index values that define the upper boundary of the chunk.


pt-table-checksum has three possible exit statuses: zero, 255, and any other value is a bitmask with flags for different problems.

A zero exit status indicates no errors, warnings, or checksum differences, or skipped chunks or tables.

A 255 exit status indicates a fatal error. In other words: the tool died or crashed. The error is printed to STDERR.

If the exit status is not zero or 255, then its value functions as a bitmask with these flags:

================  =========  ==========================================
ERROR                     1  A non-fatal error occurred
ALREADY_RUNNING           2  --pid file exists and the PID is running
NO_SLAVES_FOUND           8  No replicas or cluster nodes were found
TABLE_DIFF               16  At least one diff was found
SKIP_CHUNK               32  At least one chunk was skipped
SKIP_TABLE               64  At least one table was skipped
REPLICATION_STOPPED     128  Replica is down or stopped

If any flag is set, the exit status will be non-zero. Use the bitwise AND operation to check for a particular flag. For example, if $exit_status & 16 is true, then at least one diff was found.

As of pt-table-checksum 2.2.5, skipped chunks cause a non-zero exit status. An exit status of zero or 32 is equivalent to a zero exit status with skipped chunks in previous versions of the tool.


This tool accepts additional command-line arguments. Refer to the “SYNOPSIS” and usage information for details.


group: Connection

Prompt for a password when connecting to MySQL.


type: string

Channel name used when connected to a server using replication channels. Suppose you have two masters, master_a at port 12345, master_b at port 1236 and a slave connected to both masters using channels chan_master_a and chan_master_b. If you want to run pt-table-sync to synchronize the slave against master_a, pt-table-sync won’t be able to determine what’s the correct master since SHOW SLAVE STATUS will return 2 rows. In this case, you can use –channel=chan_master_a to specify the channel name to use in the SHOW SLAVE STATUS command.


default: yes

Check that the binlog_format is the same on all servers.

See “Replicas using row-based replication” under “LIMITATIONS”.


This option modifies the behavior of --create-replicate-table such that the replicate table’s upper and lower boundary columns are created with the BLOB data type. This is useful in cases where you have trouble checksumming tables with keys that include a binary data type or that have non-standard character sets. See --replicate.


type: time; default: 1; group: Throttle

Sleep time between checks for --max-lag.


default: yes

Check query execution plans for safety. By default, this option causes pt-table-checksum to run EXPLAIN before running queries that are meant to access a small amount of data, but which could access many rows if MySQL chooses a bad execution plan. These include the queries to determine chunk boundaries and the chunk queries themselves. If it appears that MySQL will use a bad query execution plan, the tool will skip the chunk of the table.

The tool uses several heuristics to determine whether an execution plan is bad. The first is whether EXPLAIN reports that MySQL intends to use the desired index to access the rows. If MySQL chooses a different index, the tool considers the query unsafe.

The tool also checks how much of the index MySQL reports that it will use for the query. The EXPLAIN output shows this in the key_len column. The tool remembers the largest key_len seen, and skips chunks where MySQL reports that it will use a smaller prefix of the index. This heuristic can be understood as skipping chunks that have a worse execution plan than other chunks.

The tool prints a warning the first time a chunk is skipped due to a bad execution plan in each table. Subsequent chunks are skipped silently, although you can see the count of skipped chunks in the SKIPPED column in the tool’s output.

This option adds some setup work to each table and chunk. Although the work is not intrusive for MySQL, it results in more round-trips to the server, which consumes time. Making chunks too small will cause the overhead to become relatively larger. It is therefore recommended that you not make chunks too small, because the tool may take a very long time to complete if you do.


default: yes; group: Safety

Do not checksum if any replication filters are set on any replicas. The tool looks for server options that filter replication, such as binlog_ignore_db and replicate_do_db. If it finds any such filters, it aborts with an error.

If the replicas are configured with any filtering options, you should be careful not to checksum any databases or tables that exist on the master and not the replicas. Changes to such tables might normally be skipped on the replicas because of the filtering options, but the checksum queries modify the contents of the table that stores the checksums, not the tables whose data you are checksumming. Therefore, these queries will be executed on the replica, and if the table or database you’re checksumming does not exist, the queries will cause replication to fail. For more information on replication rules, see

Replication filtering makes it impossible to be sure that the checksum queries won’t break replication (or simply fail to replicate). If you are sure that it’s OK to run the checksum queries, you can negate this option to disable the checks. See also --replicate-database.



type: string; group: Throttle

Pause checksumming until this replica’s lag is less than --max-lag. The value is a DSN that inherits properties from the master host and the connection options (--port, --user, etc.). By default, pt-table-checksum monitors lag on all connected replicas, but this option limits lag monitoring to the specified replica. This is useful if certain replicas are intentionally lagged (with pt-slave-delay for example), in which case you can specify a normal replica to monitor.



default: yes; group: Safety

Checks that tables on slaves exist and have all the checksum --columns. Tables missing on slaves or not having all the checksum --columns can cause the tool to break replication when it tries to check for differences. Only disable this check if you are aware of the risks and are sure that all tables on all slaves exist and are identical to the master.


type: string

Prefer this index for chunking tables. By default, pt-table-checksum chooses the most appropriate index for chunking. This option lets you specify the index that you prefer. If the index doesn’t exist, then pt-table-checksum will fall back to its default behavior of choosing an index. pt-table-checksum adds the index to the checksum SQL statements in a FORCE INDEX clause. Be careful when using this option; a poor choice of index could cause bad performance. This is probably best to use when you are checksumming only a single table, not an entire server.


type: int

Use only this many left-most columns of a --chunk-index. This works only for compound indexes, and is useful in cases where a bug in the MySQL query optimizer (planner) causes it to scan a large range of rows instead of using the index to locate starting and ending points precisely. This problem sometimes occurs on indexes with many columns, such as 4 or more. If this happens, the tool might print a warning related to the --[no]check-plan option. Instructing the tool to use only the first N columns of the index is a workaround for the bug in some cases.


type: size; default: 1000

Number of rows to select for each checksum query. Allowable suffixes are k, M, G. You should not use this option in most cases; prefer --chunk-time instead.

This option can override the default behavior, which is to adjust chunk size dynamically to try to make chunks run in exactly --chunk-time seconds. When this option isn’t set explicitly, its default value is used as a starting point, but after that, the tool ignores this option’s value. If you set this option explicitly, however, then it disables the dynamic adjustment behavior and tries to make all chunks exactly the specified number of rows.

There is a subtlety: if the chunk index is not unique, then it’s possible that chunks will be larger than desired. For example, if a table is chunked by an index that contains 10,000 of a given value, there is no way to write a WHERE clause that matches only 1,000 of the values, and that chunk will be at least 10,000 rows large. Such a chunk will probably be skipped because of --chunk-size-limit.

Selecting a small chunk size will cause the tool to become much slower, in part because of the setup work required for --[no]check-plan.


type: float; default: 2.0; group: Safety

Do not checksum chunks this much larger than the desired chunk size.

When a table has no unique indexes, chunk sizes can be inaccurate. This option specifies a maximum tolerable limit to the inaccuracy. The tool uses <EXPLAIN> to estimate how many rows are in the chunk. If that estimate exceeds the desired chunk size times the limit (twice as large, by default), then the tool skips the chunk.

The minimum value for this option is 1, which means that no chunk can be larger than --chunk-size. You probably don’t want to specify 1, because rows reported by EXPLAIN are estimates, which can be different from the real number of rows in the chunk. If the tool skips too many chunks because they are oversized, you might want to specify a value larger than the default of 2.

You can disable oversized chunk checking by specifying a value of 0.


type: float; default: 0.5

Adjust the chunk size dynamically so each checksum query takes this long to execute.

The tool tracks the checksum rate (rows per second) for all tables and each table individually. It uses these rates to adjust the chunk size after each checksum query, so that the next checksum query takes this amount of time (in seconds) to execute.

The algorithm is as follows: at the beginning of each table, the chunk size is initialized from the overall average rows per second since the tool began working, or the value of --chunk-size if the tool hasn’t started working yet. For each subsequent chunk of a table, the tool adjusts the chunk size to try to make queries run in the desired amount of time. It keeps an exponentially decaying moving average of queries per second, so that if the server’s performance changes due to changes in server load, the tool adapts quickly. This allows the tool to achieve predictably timed queries for each table, and for the server overall.

If this option is set to zero, the chunk size doesn’t auto-adjust, so query checksum times will vary, but query checksum sizes will not. Another way to do the same thing is to specify a value for --chunk-size explicitly, instead of leaving it at the default.


short form: -c; type: array; group: Filter

Checksum only this comma-separated list of columns. If a table doesn’t have any of the specified columns it will be skipped.

This option applies to all tables, so it really only makes sense when checksumming one table unless the tables have a common set of columns.


type: Array; group: Config

Read this comma-separated list of config files; if specified, this must be the first option on the command line.

See the --help output for a list of default config files.


default: yes

Create the --replicate database and table if they do not exist. The structure of the replicate table is the same as the suggested table mentioned in --replicate.


short form: -d; type: hash; group: Filter

Only checksum this comma-separated list of databases.


type: string; group: Filter

Only checksum databases whose names match this Perl regex. This is matched against the lowercase table name. This is the bare regex; it should not be enclosed in slashes.


short form: -F; type: string; group: Connection

Only read mysql options from the given file. You must give an absolute pathname.


Disable the QRT (Query Response Time) plugin if it is enabled.


default: yes

Delete previous checksums for each table before checksumming the table. This option does not truncate the entire table, it only deletes rows (checksums) for each table just before checksumming the table. Therefore, if checksumming stops prematurely and there was preexisting data, there will still be rows for tables that were not checksummed before the tool was stopped.

If you’re resuming from a previous checksum run, then the checksum records for the table from which the tool resumes won’t be emptied.

To empty the entire replicate table, you must manually execute TRUNCATE TABLE before running the tool.


short form: -e; type: hash; group: Filter

Only checksum tables which use these storage engines.


cumulative: yes; default: 0; group: Output

Show, but do not execute, checksum queries (disables --[no]empty-replicate-table). If specified twice, the tool actually iterates through the chunking algorithm, printing the upper and lower boundary values for each chunk, but not executing the checksum queries.


If replication is stopped, fail with an error (exit status 128) instead of waiting until replication is restarted.


type: int

Precision for FLOAT and DOUBLE number-to-string conversion. Causes FLOAT and DOUBLE values to be rounded to the specified number of digits after the decimal point, with the ROUND() function in MySQL. This can help avoid checksum mismatches due to different floating-point representations of the same values on different MySQL versions and hardware. The default is no rounding; the values are converted to strings by the CONCAT() function, and MySQL chooses the string representation. If you specify a value of 2, for example, then the values 1.008 and 1.009 will be rounded to 1.01, and will checksum as equal.


type: string

Hash function for checksums (FNV1A_64, MURMUR_HASH, SHA1, MD5, CRC32, etc).

The default is to use CRC32(), but MD5() and SHA1() also work, and you can use your own function, such as a compiled UDF, if you wish. The function you specify is run in SQL, not in Perl, so it must be available to MySQL.

MySQL doesn’t have good built-in hash functions that are fast. CRC32() is too prone to hash collisions, and MD5() and SHA1() are very CPU-intensive. The FNV1A_64() UDF that is distributed with Percona Server is a faster alternative. It is very simple to compile and install; look at the header in the source code for instructions. If it is installed, it is preferred over MD5(). You can also use the MURMUR_HASH() function if you compile and install that as a UDF; the source is also distributed with Percona Server, and it might be better than FNV1A_64().


group: Help

Show help and exit.


short form: -h; type: string; default: localhost; group: Connection

Host to connect to.


type: Hash; group: Filter

Ignore this comma-separated list of columns when calculating the checksum. If a table has all of its columns filtered by –ignore-columns, it will be skipped.


type: Hash; group: Filter

Ignore this comma-separated list of databases.


type: string; group: Filter

Ignore databases whose names match this Perl regex.


type: Hash; default: FEDERATED,MRG_MyISAM; group: Filter

Ignore this comma-separated list of storage engines.


type: Hash; group: Filter

Ignore this comma-separated list of tables. Table names may be qualified with the database name. The --replicate table is always automatically ignored.


type: string; group: Filter

Ignore tables whose names match the Perl regex. This is matched against the lowercase table name. This is the bare regex; it should not be enclosed in slashes.


type: time; default: 1s; group: Throttle

Pause checksumming until all replicas’ lag is less than this value. After each checksum query (each chunk), pt-table-checksum looks at the replication lag of all replicas to which it connects, using Seconds_Behind_Master. If any replica is lagging more than the value of this option, then pt-table-checksum will sleep for --check-interval seconds, then check all replicas again. If you specify --check-slave-lag, then the tool only examines that server for lag, not all servers.

The tool waits forever for replicas to stop lagging. If any replica is stopped, the tool waits forever until the replica is started. Checksumming continues once all replicas are running and not lagging too much.

The tool prints progress reports while waiting. If a replica is stopped, it prints a progress report immediately, then again at every progress report interval.



type: Array; default: Threads_running=25; group: Throttle

Examine SHOW GLOBAL STATUS after every chunk, and pause if any status variables are higher than the threshold. The option accepts a comma-separated list of MySQL status variables to check for a threshold. An optional =MAX_VALUE (or :MAX_VALUE) can follow each variable. If not given, the tool determines a threshold by examining the current value and increasing it by 20%.

For example, if you want the tool to pause when Threads_connected gets too high, you can specify “Threads_connected”, and the tool will check the current value when it starts working and add 20% to that value. If the current value is 100, then the tool will pause when Threads_connected exceeds 120, and resume working when it is below 120 again. If you want to specify an explicit threshold, such as 110, you can use either “Threads_connected:110” or “Threads_connected=110”.

The purpose of this option is to prevent the tool from adding too much load to the server. If the checksum queries are intrusive, or if they cause lock waits, then other queries on the server will tend to block and queue. This will typically cause Threads_running to increase, and the tool can detect that by running SHOW GLOBAL STATUS immediately after each checksum query finishes. If you specify a threshold for this variable, then you can instruct the tool to wait until queries are running normally again. This will not prevent queueing, however; it will only give the server a chance to recover from the queueing. If you notice queueing, it is best to decrease the chunk time.


short form: -p; type: string; group: Connection

Password to use when connecting. If password contains commas they must be escaped with a backslash: “exam,ple”


type: string

Execution will be paused while the file specified by this param exists.


type: string

Create the given PID file. The tool won’t start if the PID file already exists and the PID it contains is different than the current PID. However, if the PID file exists and the PID it contains is no longer running, the tool will overwrite the PID file with the current PID. The PID file is removed automatically when the tool exits.


type: string

Perl module file that defines a pt_table_checksum_plugin class. A plugin allows you to write a Perl module that can hook into many parts of pt-table-checksum. This requires a good knowledge of Perl and Percona Toolkit conventions, which are beyond this scope of this documentation. Please contact Percona if you have questions or need help.

See “PLUGIN” for more information.


short form: -P; type: int; group: Connection

Port number to use for connection.


type: array; default: time,30

Print progress reports to STDERR.

The value is a comma-separated list with two parts. The first part can be percentage, time, or iterations; the second part specifies how often an update should be printed, in percentage, seconds, or number of iterations. The tool prints progress reports for a variety of time-consuming operations, including waiting for replicas to catch up if they become lagged.


short form: -q; cumulative: yes; default: 0

Print only the most important information (disables --progress). Specifying this option once causes the tool to print only errors, warnings, and tables that have checksum differences.

Specifying this option twice causes the tool to print only errors. In this case, you can use the tool’s exit status to determine if there were any warnings or checksum differences.


type: int

Number of levels to recurse in the hierarchy when discovering replicas. Default is infinite. See also --recursion-method and “REPLICA CHECKS”.


type: array; default: processlist,hosts

Preferred recursion method for discovering replicas. pt-table-checksum performs several “REPLICA CHECKS” before and while running.

Although replicas are not required to run pt-table-checksum, the tool cannot detect diffs on slaves that it cannot discover. Therefore, a warning is printed and the “EXIT STATUS” is non-zero if no replicas are found and the method is not none. If this happens, try a different recursion method, or use the dsn method to specify the replicas to check.

Possible methods are:

===========  =============================================
hosts        SHOW SLAVE HOSTS
cluster      SHOW STATUS LIKE 'wsrep\_incoming\_addresses'
dsn=DSN      DSNs from a table
none         Do not find slaves

The processlist method is the default, because SHOW SLAVE HOSTS is not reliable. However, if the server uses a non-standard port (not 3306), then the hosts method becomes the default because it works better in this case.

The hosts method requires replicas to be configured with report_host, report_port, etc.

The cluster method requires a cluster based on Galera 23.7.3 or newer, such as Percona XtraDB Cluster versions 5.5.29 and above. This will auto-discover nodes in a cluster using SHOW STATUS LIKE 'wsrep\_incoming\_addresses'. You can combine cluster with processlist and hosts to auto-discover cluster nodes and replicas, but this functionality is experimental.

The dsn method is special: rather than automatically discovering replicas, this method specifies a table with replica DSNs. The tool will only connect to these replicas. This method works best when replicas do not use the same MySQL username or password as the master, or when you want to prevent the tool from connecting to certain replicas. The dsn method is specified like: --recursion-method dsn=h=host,D=percona,t=dsns. The specified DSN must have D and t parts, or just a database-qualified t part, which specify the DSN table. The DSN table must have the following structure:

  `parent_id` int(11) DEFAULT NULL,
  `dsn` varchar(255) NOT NULL,
  PRIMARY KEY (`id`)

DSNs are ordered by id, but id and parent_id are otherwise ignored. The dsn column contains a replica DSN like it would be given on the command line, for example: "h=replica_host,u=repl_user,p=repl_pass".

The none method makes the tool ignore all slaves and cluster nodes. This method is not recommended because it effectively disables the “REPLICA CHECKS” and no differences can be found. It is useful, however, if you only need to write checksums on the master or a single cluster node. The safer alternative is --no-replicate-check: the tool finds replicas and cluster nodes, performs the “REPLICA CHECKS”, but does not check for differences. See --[no]replicate-check.


type: string; default: percona.checksums

Write checksum results to this table. The replicate table must have this structure (MAGIC_create_replicate):

CREATE TABLE checksums (
   db             CHAR(64)     NOT NULL,
   tbl            CHAR(64)     NOT NULL,
   chunk          INT          NOT NULL,
   chunk_time     FLOAT            NULL,
   chunk_index    VARCHAR(200)     NULL,
   lower_boundary TEXT             NULL,
   upper_boundary TEXT             NULL,
   this_crc       CHAR(40)     NOT NULL,
   this_cnt       INT          NOT NULL,
   master_crc     CHAR(40)         NULL,
   master_cnt     INT              NULL,
   PRIMARY KEY (db, tbl, chunk),
   INDEX ts_db_tbl (ts, db, tbl)

Note: lower_boundary and upper_boundary data type can be BLOB. See --binary-index.

By default, --create-replicate-table is true, so the database and the table specified by this option are created automatically if they do not exist.

Be sure to choose an appropriate storage engine for the replicate table. If you are checksumming InnoDB tables, and you use MyISAM for this table, a deadlock will break replication, because the mixture of transactional and non-transactional tables in the checksum statements will cause it to be written to the binlog even though it had an error. It will then replay without a deadlock on the replicas, and break replication with “different error on master and slave.” This is not a problem with pt-table-checksum; it’s a problem with MySQL replication, and you can read more about it in the MySQL manual.

The replicate table is never checksummed (the tool automatically adds this table to --ignore-tables).


default: yes

Check replicas for data differences after finishing each table. The tool finds differences by executing a simple SELECT statement on all detected replicas. The query compares the replica’s checksum results to the master’s checksum results. It reports differences in the DIFFS column of the output.


Check replicas for consistency without executing checksum queries. This option is used only with --[no]replicate-check. If specified, pt-table-checksum doesn’t checksum any tables. It checks replicas for differences found by previous checksumming, and then exits. It might be useful if you run pt-table-checksum quietly in a cron job, for example, and later want a report on the results of the cron job, perhaps to implement a Nagios check.


type: int; default: 1

Retry checksum comparison this many times when a difference is encountered. Only when a difference persists after this number of checks is it considered valid. Using this option with a value of 2 or more alleviates spurious differences that arise when using the –resume option.


type: string

USE only this database. By default, pt-table-checksum executes USE to select the database that contains the table it’s currently working on. This is is a best effort to avoid problems with replication filters such as binlog_ignore_db and replicate_ignore_db. However, replication filters can create a situation where there simply is no one right way to do things. Some statements might not be replicated, and others might cause replication to fail. In such cases, you can use this option to specify a default database that pt-table-checksum selects with USE, and never changes. See also --[no]check-replication-filters.


Resume checksumming from the last completed chunk (disables --[no]empty-replicate-table). If the tool stops before it checksums all tables, this option makes checksumming resume from the last chunk of the last table that it finished.


type: int; default: 2

Retry a chunk this many times when there is a nonfatal error. Nonfatal errors are problems such as a lock wait timeout or the query being killed.


type: time

How long to run. Default is to run until all tables have been checksummed. These time value suffixes are allowed: s (seconds), m (minutes), h (hours), and d (days). Combine this option with --resume to checksum as many tables within an allotted time, resuming from where the tool left off next time it is ran.


type: string; default: #

The separator character used for CONCAT_WS(). This character is used to join the values of columns when checksumming.


type: DSN; repeatable: yes

DSN to skip when checking slave lag. It can be used multiple times. Example: –skip-check-slave-lag h=127.1,P=12345 –skip-check-slave-lag h=127.1,P=12346


type: string

Sets the user to be used to connect to the slaves. This parameter allows you to have a different user with less privileges on the slaves but that user must exist on all slaves.


type: string

Sets the password to be used to connect to the slaves. It can be used with –slave-user and the password for the user must be the same on all slaves.


type: Array; group: Connection

Set the MySQL variables in this comma-separated list of variable=value pairs.

By default, the tool sets:


Variables specified on the command line override these defaults. For example, specifying --set-vars wait_timeout=500 overrides the defaultvalue of 10000.

The tool prints a warning and continues if a variable cannot be set.


short form: -S; type: string; group: Connection

Socket file to use for connection.


type: float; default: 1.0

When a master table is marked to be checksummed in only one chunk but a slave table exceeds the maximum accepted size for this, the table is skipped. Since number of rows are often rough estimates, many times tables are skipped needlessly for very small differences. This option provides a max row excess tolerance to prevent this. For example a value of 1.2 will tolerate slave tables with up to 20% excess rows.


short form: -t; type: hash; group: Filter

Checksum only this comma-separated list of tables. Table names may be qualified with the database name.


type: string; group: Filter

Checksum only tables whose names match this Perl regex.


Add TRIM() to VARCHAR columns (helps when comparing 4.1 to >= 5.0). This is useful when you don’t care about the trailing space differences between MySQL versions that vary in their handling of trailing spaces. MySQL 5.0 and later all retain trailing spaces in VARCHAR, while previous versions would remove them. These differences will cause false checksum differences.


Truncate the replicate table before starting the checksum. This parameter differs from –empty-replicate-table which only deletes the rows for the table being checksummed when starting the checksum for that table, while –truncate-replicate-table will truncate the replicate table at the beginning of the process and thus, all previous checksum information will be losti, even if the process stops due to an error.


short form: -u; type: string; group: Connection

User for login if not current user.


group: Help

Show version and exit.


default: yes

Check for the latest version of Percona Toolkit, MySQL, and other programs.

This is a standard “check for updates automatically” feature, with two additional features. First, the tool checks its own version and also the versions of the following software: operating system, Percona Monitoring and Management (PMM), MySQL, Perl, MySQL driver for Perl (DBD::mysql), and Percona Toolkit. Second, it checks for and warns about versions with known problems. For example, MySQL 5.5.25 had a critical bug and was re-released as 5.5.25a.

A secure connection to Percona’s Version Check database server is done to perform these checks. Each request is logged by the server, including software version numbers and unique ID of the checked system. The ID is generated by the Percona Toolkit installation script or when the Version Check database call is done for the first time.

Any updates or known problems are printed to STDOUT before the tool’s normal output. This feature should never interfere with the normal operation of the tool.

For more information, visit


type: string

Do only rows matching this WHERE clause. You can use this option to limit the checksum to only part of the table. This is particularly useful if you have append-only tables and don’t want to constantly re-check all rows; you could run a daily job to just check yesterday’s rows, for instance.

This option is much like the -w option to mysqldump. Do not specify the WHERE keyword. You might need to quote the value. Here is an example:

:program:`pt-table-checksum` --where "ts > CURRENT_DATE - INTERVAL 1 DAY"


By default, pt-table-checksum attempts to find and connect to all replicas connected to the master host. This automated process is called “slave recursion” and is controlled by the --recursion-method and --recurse options. The tool performs these checks on all replicas:

  1. --[no]check-replication-filters

pt-table-checksum checks for replication filters on all replicas because they can complicate or break the checksum process. By default, the tool will exit if any replication filters are found, but this check can be disabled by specifying --no-check-replication-filters.

  1. --replicate table

pt-table-checksum checks that the --replicate table exists on all replicas, else checksumming can break replication when updates to the table on the master replicate to a replica that doesn’t have the table. This check cannot be disabled, and the tool waits forever until the table exists on all replicas, printing --progress messages while it waits.

  1. Single chunk size

If a table can be checksummed in a single chunk on the master, pt-table-checksum will check that the table size on all replicas is less than --chunk-size * --chunk-size-limit. This prevents a rare problem where the table on the master is empty or small, but on a replica it is much larger. In this case, the single chunk checksum on the master would overload the replica.

Another rare problem occurs when the table size on a replica is close to --chunk-size * --chunk-size-limit. In such cases, the table is more likely to be skipped even though it’s safe to checksum in a single chunk. This happens because table sizes are estimates. When those estimates and --chunk-size * --chunk-size-limit are almost equal, this check becomes more sensitive to the estimates’ margin of error rather than actual significant differences in table sizes. Specifying a larger value for --chunk-size-limit helps avoid this problem.

This check cannot be disabled.

  1. Lag

After each chunk, pt-table-checksum checks the lag on all replicas, or only the replica specified by --check-slave-lag. This helps the tool not to overload the replicas with checksum data. There is no way to disable this check, but you can specify a single replica to check with --check-slave-lag, and if that replica is the fastest, it will help prevent the tool from waiting too long for replica lag to abate.

  1. Checksum chunks

When pt-table-checksum finishes checksumming a table, it waits for the last checksum chunk to replicate to all replicas so it can perform the --[no]replicate-check. Disabling that option by specifying –no-replicate-check disables this check, but it also disables immediate reporting of checksum differences, thereby requiring a second run of the tool with --replicate-check-only to find and print checksum differences.


The file specified by --plugin must define a class (i.e. a package) called pt_table_checksum_plugin with a new() subroutine. The tool will create an instance of this class and call any hooks that it defines. No hooks are required, but a plugin isn’t very useful without them.

These hooks, in this order, are called if defined:


Each hook is passed different arguments. To see which arguments are passed to a hook, search for the hook’s name in the tool’s source code, like:

# --plugin hook
if ( $plugin && $plugin->can('init') ) {
      slaves         => $slaves,
      slave_lag_cxns => $slave_lag_cxns,
      repl_table     => $repl_table,

The comment # --plugin hook precedes every hook call.

Please contact Percona if you have questions or need help.


These DSN options are used to create a DSN. Each option is given like option=value. The options are case-sensitive, so P and p are not the same option. There cannot be whitespace before or after the = and if the value contains whitespace it must be quoted. DSN options are comma-separated. See the percona-toolkit manpage for full details.

  • A

dsn: charset; copy: yes

Default character set.

  • D

copy: no

DSN table database.

  • F

dsn: mysql_read_default_file; copy: yes

Defaults file for connection values.

  • h

dsn: host; copy: yes

Connect to host.

  • p

dsn: password; copy: yes

Password to use when connecting. If password contains commas they must be escaped with a backslash: “exam,ple”

  • P

dsn: port; copy: yes

Port number to use for connection.

  • S

dsn: mysql_socket; copy: no

Socket file to use for connection.

  • t

copy: no

DSN table table.

  • u

dsn: user; copy: yes

User for login if not current user.


The environment variable PTDEBUG enables verbose debugging output to STDERR. To enable debugging and capture all output to a file, run the tool like:

PTDEBUG=1 pt-table-checksum ... > FILE 2>&1

Be careful: debugging output is voluminous and can generate several megabytes of output.


Using <PTDEBUG> might expose passwords. When debug is enabled, all command line parameters are shown in the output.


You need Perl, DBI, DBD::mysql, and some core packages that ought to be installed in any reasonably new version of Perl.


For a list of known bugs, see

Please report bugs at Include the following information in your bug report:

  • Complete command-line used to run the tool

  • Tool --version

  • MySQL version of all servers involved

  • Output from the tool including STDERR

  • Input files (log/dump/config files, etc.)

If possible, include debugging output by running the tool with PTDEBUG; see “ENVIRONMENT”.


Visit to download the latest release of Percona Toolkit. Or, get the latest release from the command line:




You can also get individual tools from the latest release:


Replace TOOL with the name of any tool.


Baron Schwartz and Daniel Nichter


Claus Jeppesen, Francois Saint-Jacques, Giuseppe Maxia, Heikki Tuuri, James Briggs, Martin Friebe, and Sergey Zhuravlev


This tool is part of Percona Toolkit, a collection of advanced command-line tools for MySQL developed by Percona. Percona Toolkit was forked from two projects in June, 2011: Maatkit and Aspersa. Those projects were created by Baron Schwartz and primarily developed by him and Daniel Nichter. Visit to learn about other free, open-source software from Percona.


pt-table-checksum 3.5.7