TokuDB has been deprecated by its upstream maintainer. It is disabled from MariaDB 10.5 and has been removed in MariaDB 10.6 - MDEV-19780 . We recommend MyRocks as a long-term migration path.
This page lists system variables that are related to TokuDB.
See Server System Variables for a complete list of system variables and instructions on setting them, and Full list of MariaDB options, system and status variables for a complete list of all options, statis variable and system variables in MariaDB.
Description: Print errors for alter table operations.
Description: Time in seconds that ANALYZE operations spend on each index when calculating cardinality. Accurate cardinality helps in particular with the performance of complex queries. If no analyzes are run, cardinality are 1 for primary indexes, and unknown (NULL) for other types of indexes.
Description: Uncompressed size of internal fractal tree and leaf nodes. Changing will only affect tables created after the new setting is in effect. Existing tables will keep the setting they were created with unless the table is dumped and reloaded.
Default Value: 4194304
(4MB)
Range: 0
to 18446744073709551615
Description: If set to 1
(the default), the bulk fetch algorithm is used for SELECT's and DELETE's, including related statements such as INSERT INTO.
Description: Size in bytes of the TokuDB cache. This variable is read-only and cannot be changed dynamically. To change the value, either set the value in the my.cnf
file prior to loading TokuDB or restart MariaDB after modifying the configuration. If you have loaded the plugin but not used TokuDB yet, you can unload the plugin then reload it and MariaDB will reload the plugin with the setting from the configuration file. Setting to at least half of the available memory is recommended, although if using directIO instead of buffered IO (see tokudb_directio) , up to 80% of the available memory is recommended. Decrease if other applications require significant memory or swapping is degrading performance.
Default Value: Half of the total system memory
Description: Check if jemalloc is linked.
Description: Mechanism to lock out TokuDB checkpoints. When set to 1
, TokuDB checkpoints are locked out. Setting to 0
, or disconnecting the client, releases the lock.
tokudb_checkpoint_on_flush_logs
Description: TokuDB checkpoint on flush logs.
tokudb_checkpointing_period
Description: Time in seconds between the beginning of each checkpoint. It is recommended to leave this at the default setting of 1 minute.
tokudb_cleaner_iterations
Description: Number of internal nodes processed in each cleaner thread period (see tokudb_cleaner_period). Setting to 0
turns off cleaner threads.
Range: 0
to 18446744073709551615
Description: Frequency in seconds for the running of the cleaner thread. Setting to 0
turns off cleaner threads.
Range: 0
to 18446744073709551615
Description: Whether or not the transaction log is flushed upon transaction commit. Flushing has a minor performance penalty, but switching it off means that committed transactions may not survive a server crash.
tokudb_create_index_online
Description: Whether indexes are hot or not. Hot, or online, indexes (the default) mean that the table is available for inserting and updates while the index is being created. It is slower to create hot indexes.
Description: Directory where the TokuDB data is stored. By default the variable is empty, in which case the regular datadir is used.
Default Value: Empty (the MariaDB datadir is used)
Description: Setting to a non-zero value turns on various TokuDB debug traces.
Range: 0
to 18446744073709551615
Description: When set to ON, TokuDB writes use Direct IO instead of Buffered IO. tokudb_cache_size should be adjusted when using DirectIO.
Description: If set to ON
(OFF
is default), hot alter table is disabled.
tokudb_disable_prefetching
Description: If prefetching is not disabled (the default), range queries usually benefit from aggressive prefetching of blocks of rows. For range queries with LIMIT clauses, this can create unnecessary IO, and so prefetching can be disabled if these make up a majority of range queries.
tokudb_disable_slow_alter
Description: Usually, TokuDB permits column addition, deletion, expansion, and renaming with minimal locking, very quickly. This variable determines whether certain slow [alter|ALTER]] table statements that cannot take advantage of this feature are permitted. Statements that are slow are those that include a mix of column additions, deletions or expansions, for example, ALTER TABLE t1 ADD COLUMN c1 int, DROP COLUMN c2
.
Description: TokuDB algorithm to check if the table is empty when opened. Setting to disabled
will reduce this overhead.
Valid Values: lr
, rl
, disabled
tokudb_fs_reserve_percent
Description: If this percentage of the filesystem is not free, inserts are prohibited. Recommended value is half the size of the available memory. Once disabled, inserts are re-enabled once twice the reserve is available. TokuDB will freeze entirely if the disk becomes entirely full.
Description: fsync() operations frequency in milliseconds. If set to 0
, the default, tokudb_commit_sync control fsync() behavior.
Range: 0
to 18446744073709551615
Warning: currently values in the 1000-2000 range seem to cause server crashes, see MDEV-16732
tokudb_hide_default_row_format
Description: Hide the default row format.
Description: Control lock tree kill callback frequency.
Range: 0
to 18446744073709551615
Description: Empty by default, when a lock deadlock is detected, or a lock request times out, set to a JSON document describing the most recent lock conflict. Only set when the first bit of tokudb_lock_timeout_debug is set.
Description: If set to 1
, the default, bulk loader intermediate data is compressed, otherwise it is uncompressed. Also see tokudb_tmp_dir.
tokudb_loader_memory_size
Description: Memory limit for each loader instance used by the TokuDB bulk loader. Memory is taken from the TokuDB cache (tokudb_cache_size), so current cache data may need to be cleared for the loader to begin. Increase if tables are very larger, with multiple secondary indexes.
Default Value: 100000000
(100M)
Range: 0
to 18446744073709551615
Description: Time in milliseconds that a transaction will wait for a lock held by another transaction to be released before timing out with a lock wait timeout
error (-30994). Setting to 0
disables lock waits.
Default Value: 4000
(4 seconds)
Range: 0
to 18446744073709551615
tokudb_lock_timeout_debug
Description: When bit zero is set (default 1
), a JSON document describing the most recent lock conflict is reported to tokudb_last_lock_timeout. When set to 0
, no lock conflicts are reported. When bit one is set, the JSON document is printed to the error log.
Description: Directory where the TokuDB log files are stored. By default the variable is empty, in which case the regular datadir is used.
Default Value: Empty (the MariaDB datadir is used)
Description: Max memory for locks.
tokudb_optimize_index_fraction
Description: When deleting a percentage of the tree (useful when the left side of the tree has many deletions, such as a pattern with increasing ids or dates), it's possible to optimize a subset of the fractal tree, as determined by the value of this variable, which ranges from 0.0
to 1.0
(indicating the whole tree).
tokudb_optimize_index_name
Description: If set to an index name, will optimize that single index in a table. Empty by default.
Description: Table optimization utilizes all available resources by default. This variable allows the table optimization speed to be limited in order to reduce the overall resources used. The limit places an upper bound on the number of fractal tree leaf nodes that are optimized per second. 0
, the default, imposes no limit.
Range: 0
to 18446744073709551615
Description: Mode for primary key inserts using either REPLACE INTO or INSERT IGNORE on tables with no secondary index, or where all columns in the secondary index are in the primary key. For example PRIMARY KEY (a,b,c), key (b,c)
1
: Fast inserts if no triggers are defined, otherwise inserts may be slow. Row-based replication will not work.
2
: Slow inserts. Triggers and row-based replication work normally.
Description: If set to 0
(1
is default), fast bulk loading are switched off. Usually, TokuDB obtains a table lock on empty tables. If, as is usual, only one transaction is loading the table, this speeds up the inserts. However, if many transactions are loading, only one can have access at a time, so setting this to 0
, avoiding the lock, will speed inserts up in that situation.
Description: Uncompressed size in bytes of the read blocks of the fractal tree leaves. Changing will only affect tables created after the new setting is in effect. Existing tables will keep the setting they were created with unless the table is dumped and reloaded. Larger values are better for large range scans and higher compressions, while smaller values are better for point and small range scans.
Default Value: 65536
(64KB)
Range: 4096
to 4294967295
Description: Per-client size in bytes of the buffer used for storing bulk fetched values as part of a large range query. Reduce if there are many simultaneous clients. Setting to 0
disables bulk fetching.
Default Value: 131072
(128KB)
tokudb_read_status_frequency
Description: Progress is measured every this many reads for display by SHOW PROCESSLIST . Useful to set to 1
to examine slow queries.
Description: Compression algorithm used by default to compress data. Can be overridden by a row format specified in the CREATE TABLE statement. note that the library can be specified directly, or an alias used, the mapping of which may change in future. Note that in MariaDB 5.5, and before MariaDB 10.0.10, the compression type did not default to this value. See TokuDB Differences.
tokudb_default
, tokudb_zlib
: Use the zlib library,
tokudb_fast
, tokudb_quicklz
: Use the quicklz library, the lightest compression with low CPU usage,
tokudb_small
, tokudb_lzma
: Use the lzma library. the highest compression and highest CPU usage
tokudb_uncompressed
: No compression is used.
Default Value: tokudb_zlib
Valid Values: tokudb_default
, tokudb_fast
, tokudb_small
, tokudb_zlib
, tokudb_quicklz
, tokudb_lzma
, tokudb_uncompressed
tokudb_rpl_check_readonly
Description: By default, when the slave is in read only mode, row events are run from the binary log using TokuDB's read-free replication (RFR). Setting this variable to OFF
turns off the slave read only check, allowing RFR to run when the slave is not read-only. Be careful that you understand the consequences if setting this variable.
Description: If set to OFF
(ON
is default), and binlog_format to ROW
and read_only to ON
, TokuDB replication slaves will not perform row lookups for update or delete row log events, removing the need for the associated IO.
tokudb_rpl_lookup_rows_delay
Description: Can be used to simulate long disk reads by sleeping for the specified time, in microseconds, before the row lookup query. Only useful to change in a test environment.
Description: If set to OFF
(ON
is default), and binlog_format to ROW
and read_only to ON
, TokuDB replication slaves will skip uniqueness checks on inserts and updates, removing the associated IO.
tokudb_rpl_unique_checks_delay
Description: Can be used to simulate long disk reads by sleeping for the specified time, in microseconds, before the row lookup query. Only useful to change in a test environment.
Description: Whether or not the prepare phase of an XA transaction performs an fsync().
Description: Directory where the TokuDB bulk loaders temporary files are stored. Can be very large, and useful to place on a separate disk. By default the variable is empty, in which case the regular datadir is used. tokudb_load_save_space determines whether the data is compressed or not. The error message ERROR 1030 (HY000): Got error 1 from storage engine
could indicate that the disk has run out of space.
Data Type: directory name
Default Value: Empty (the MariaDB datadir is used)
Description: The TokuDB version of the plugin included on MariaDB.
tokudb_write_status_frequency
Description: Progress is measured every this many writes for display by SHOW PROCESSLIST. Useful to set to 1
to examine slow queries.
This page is licensed: CC BY-SA / Gnu FDL
RetroSearch is an open source project built by @garambo | Open a GitHub Issue
Search and Browse the WWW like it's 1997 | Search results from DuckDuckGo
HTML:
3.2
| Encoding:
UTF-8
| Version:
0.7.4