This feature adds microsecond time resolution and additional statistics to the slow query log output. It lets you enable or disable the slow query log at runtime, adds logging for the slave SQL thread, and adds fine-grained control over what and how much to log into the slow query log.
You can use Percona-Toolkit‘s pt-query-digest tool to aggregate similar queries together and report on those that consume the most execution time.
- Author / Origin: Maciej Dobrzanski
Filters the slow log by the query’s execution plan. The value is a comma-delimited string, and can contain any combination of the following values:
- qc_miss: The query was not found in the query cache.
- full_scan: The query performed a full table scan.
- full_join: The query performed a full join (a join without indexes).
- tmp_table: The query created an implicit internal temporary table.
- tmp_table_on_disk: The query’s temporary table was stored on disk.
- filesort: The query used a filesort.
- filesort_on_disk: The filesort was performed on disk.
Values are OR’ed together. If the string is empty, then the filter is disabled. If it is not empty, then queries will only be logged to the slow log if their execution plan matches one of the types of plans present in the filter.
For example, to log only queries that perform a full table scan, set the value to full_scan. To log only queries that use on-disk temporary storage for intermediate results, set the value to tmp_table_on_disk,filesort_on_disk.
Specifies semantic of log_slow_rate_limit - session or query.
Behavior of this variable depends from log_slow_rate_type.
Specifies that only a fraction of session/query should be logged. Logging is enabled for every nth session/query. By default, n is 1, so logging is enabled for every session/query. Please note: when log_slow_rate_type is session rate limiting is disabled for the replication thread.
Note that every query has global unique query_id and every connection can has it own (session) log_slow_rate_limit. Decision “log or no” calculated in following manner:
- if log_slow_rate_limit is 1 - log every query
- If log_slow_rate_limit > 1 - randomly log every 1/log_slow_rate_limit query.
This allows flexible setup logging behavior.
For example, if you set the value to 100, then one percent of sessions/queries will be logged. In Percona Server 5.6.13-60.6 information about the log_slow_rate_limit has been added to the slow query log. This means that if the log_slow_rate_limit is effective it will be reflected in the slow query log for each written query. Example of the output looks like this:
Log_slow_rate_type: query Log_slow_rate_limit: 10
Prior to 5.6.17-65.0 implementation of the log_slow_rate_type set to query with log_slow_rate_limit feature would log every nth query deterministically. With the current implementation each query has a non-deterministic probability of 1/n to get logged.
If TRUE, statements executed by stored procedures are logged to the slow if it is open.
Support for logging stored procedures doesn’t involve triggers, so they won’t be logged even if this feature is enabled.
Specifies how much information to include in your slow log. The value is a comma-delimited string, and can contain any combination of the following values:
- microtime: Log queries with microsecond precision.
- query_plan: Log information about the query’s execution plan.
- innodb: Log InnoDB statistics.
- minimal: Equivalent to enabling just microtime.
- standard: Equivalent to enabling microtime,innodb.
- full: Equivalent to all other values OR’ed together without the profiling and profiling_use_getrusage options.
- profiling: Enables profiling of all queries in all connections.
- profiling_use_getrusage: Enables usage of the getrusage function.
Values are OR’ed together.
For example, to enable microsecond query timing and InnoDB statistics, set this option to microtime,innodb or standard. To turn all options on, set the option to full.
If TRUE, a timestamp is printed on every slow log record. Multiple records may have the same time.
Normally, entries to the slow query log are in seconds precision, in this format:
# Time: 090402 9:23:36 # User@Host: XXX @ XXX [10.X.X.X]
If slow_query_log_timestamp_precision =microsecond, entries to the slow query log are in microsecond precision, in this format:
# Time: 090402 9:23:36.123456 # User@Host: XXX @ XXX [10.X.X.X]
Specifies which variables have global scope instead of local. Value is a “flag” variable - you can specify multiple values separated by commas
- none: All variables use local scope
- log_slow_filter: Global variable log_slow_filter has effect (instead of local)
- log_slow_rate_limit: Global variable log_slow_rate_limit has effect (instead of local)
- log_slow_verbosity: Global variable log_slow_verbosity has effect (instead of local)
- long_query_time: Global variable long_query_time has effect (instead of local)
- min_examined_row_limit: Global variable min_examined_row_limit has effect (instead of local)
- all Global variables has effect (instead of local)
This variable can be used to specify the query execution time after which the query will be written to the slow query log. It can be used to specify an additional execution time threshold for the slow query log, that, when exceeded, will cause a query to be logged unconditionally, that is, log_slow_rate_limit will not apply to it.
The feature adds more information to the slow log output. Here is a sample log entry:
# Time: 130601 8:01:06.058915 # User@Host: root[root] @ localhost  Id: 42 # Schema: imdb Last_errno: 0 Killed: 0 # Query_time: 7.725616 Lock_time: 0.000328 Rows_sent: 4 Rows_examined: 1543720 Rows_affected: 0 # Bytes_sent: 272 Tmp_tables: 0 Tmp_disk_tables: 0 Tmp_table_sizes: 0 # QC_Hit: No Full_scan: Yes Full_join: No Tmp_table: No Tmp_table_on_disk: No # Filesort: No Filesort_on_disk: No Merge_passes: 0 SET timestamp=1370073666; SELECT id,title,production_year FROM title WHERE title = 'Bambi';
Another example (log_slow_verbosity =profiling):
# Time: 130601 8:03:20.700441 # User@Host: root[root] @ localhost  Id: 43 # Schema: imdb Last_errno: 0 Killed: 0 # Query_time: 7.815071 Lock_time: 0.000261 Rows_sent: 4 Rows_examined: 1543720 Rows_affected: 0 # Bytes_sent: 272 # Profile_starting: 0.000125 Profile_starting_cpu: 0.000120 Profile_checking_permissions: 0.000021 Profile_checking_permissions_cpu: 0.000021 Profile_Opening_tables: 0.000049 Profile_Opening_tables_cpu: 0.000048 Profile_init: 0.000048 Profile_init_cpu: 0.000049 Profile_System_lock: 0.000049 Profile_System_lock_cpu: 0.000048 Profile_optimizing: 0.000024 Profile_optimizing_cpu: 0.000024 Profile_statistics: 0.000036 Profile_statistics_cpu: 0.000037 Profile_preparing: 0.000029 Profile_preparing_cpu: 0.000029 Profile_executing: 0.000012 Profile_executing_cpu: 0.000012 Profile_Sending_data: 7.814583 Profile_Sending_data_cpu: 7.811634 Profile_end: 0.000013 Profile_end_cpu: 0.000012 Profile_query_end: 0.000014 Profile_query_end_cpu: 0.000014 Profile_closing_tables: 0.000023 Profile_closing_tables_cpu: 0.000023 Profile_freeing_items: 0.000051 Profile_freeing_items_cpu: 0.000050 Profile_logging_slow_query: 0.000006 Profile_logging_slow_query_cpu: 0.000006 # Profile_total: 7.815085 Profile_total_cpu: 7.812127 SET timestamp=1370073800; SELECT id,title,production_year FROM title WHERE title = 'Bambi';
Each slow log entry now contains a connection identifier, so you can trace all the queries coming from a single connection. This is the same value that is shown in the Id column in SHOW FULL PROCESSLIST or returned from the CONNECTION_ID() function.
Each entry also contains a schema name, so you can trace all the queries whose default database was set to a particular schema.
# Id: 43 Schema: imdb
This is the original functionality offered by the microslow feature. Query_time and Lock_time are logged with microsecond resolution.
The feature also adds information about how many rows were examined for SELECT queries, and how many were analyzed and affected for UPDATE, DELETE, and INSERT queries,
# Query_time: 0.962742 Lock_time: 0.000202 Rows_sent: 4 Rows_examined: 1543719 Rows_affected: 0
Values and context:
- Rows_examined: Number of rows scanned - SELECT
- Rows_affected: Number of rows changed - UPDATE, DELETE, INSERT
The feature provides information about the amount of bytes sent for the result of the query and the number of temporary tables created for its execution - differentiated by whether they were created on memory or on disk - with the total number of bytes used by them.
# Bytes_sent: 8053 Tmp_tables: 1 Tmp_disk_tables: 0 Tmp_table_sizes: 950528
Values and context:
- Bytes_sent: The amount of bytes sent for the result of the query
- Tmp_tables: Number of temporary tables created on memory for the query
- Tmp_disk_tables: Number of temporary tables created on disk for the query
- Tmp_table_sizes: Total Size in bytes for all temporary tables used in the query
Each query can be executed in various ways. For example, it may use indexes or do a full table scan, or a temporary table may be needed. These are the things that you can usually see by running EXPLAIN on the query. The feature will now allow you to see the most important facts about the execution in the log file.
# QC_Hit: No Full_scan: Yes Full_join: No Tmp_table: No Tmp_table_on_disk: No # Filesort: No Filesort_on_disk: No Merge_passes: 0
The values and their meanings are documented with the log_slow_filter option.
The final part of the output is the InnoDB usage statistics. MySQL currently shows many per-session statistics for operations with SHOW SESSION STATUS, but that does not include those of InnoDB, which are always global and shared by all threads. This feature lets you see those values for a given query.
# InnoDB_IO_r_ops: 6415 InnoDB_IO_r_bytes: 105103360 InnoDB_IO_r_wait: 0.001279 # InnoDB_rec_lock_wait: 0.000000 InnoDB_queue_wait: 0.000000 # InnoDB_pages_distinct: 6430
- innodb_IO_r_ops: Counts the number of page read operations scheduled. The actual number of read operations may be different, but since this can be done asynchronously, there is no good way to measure it.
- innodb_IO_r_bytes: Similar to innodb_IO_r_ops, but the unit is bytes.
- innodb_IO_r_wait: Shows how long (in seconds) it took InnoDB to actually read the data from storage.
- innodb_rec_lock_wait: Shows how long (in seconds) the query waited for row locks.
- innodb_queue_wait: Shows how long (in seconds) the query spent either waiting to enter the InnoDB queue or inside that queue waiting for execution.
- innodb_pages_distinct: Counts approximately the number of unique pages the query accessed. The approximation is based on a small hash array representing the entire buffer pool, because it could take a lot of memory to map all the pages. The inaccuracy grows with the number of pages accessed by a query, because there is a higher probability of hash collisions.
If the query did not use InnoDB tables, that information is written into the log instead of the above statistics.