Version 2.0.0 (2023-03-20)¶
Release date | March 20, 2023 |
---|---|
Installation | Install pg_stat_monitor |
Release highlights¶
pg_stat_monitor
2.0.0 is Generally Available with Percona Distribution for PostgreSQL, enabling you to use it in production environments. Its improved internal architecture leads to fewer lock acquisitions, and therefore an improved performance by approximately 20% when tested usingpgbench
.-
Added support for PostgreSQL 15 by adding the following columns to
pg_stat_monitor
view:temp_blk_read_time
temp_blk_write_time
jit_functions
jit_generation_time
jit_inlining_count
jit_inlining_time
jit_optimization_count
jit_optimization_time
jit_emission_count
jit_emission_time
-
Improved consistency with
pg_stat_statements
- Nowpg_stat_monitor
view has exactly the same column names, columns and data types as thepg_stat_statements
view for every major version of PostgreSQL from 11 to 15. - Histogram output now allows users to see the whole data range, including the queries with the execution time lower than the minimum time, and exceeding the maximum time specified by the user.
- The query text is now stored in memory, improving the memory buffer management and preventing its overflow
- CPU and user time are now captured for utility statements, helping you understand the resource consumption for these statements execution.
- PMM integration improvements:
- An indication of buckets as “done” vs “current” eliminates the need for the tool to evaluate the bucket status and facilitates the accurate data display
- The generation of a unique ID of a query, independent from PostgreSQL versions, database, users or schemas, improves observability of query planning, execution and performance
Incompatibility with previous releases:¶
- The
pg_stat_monitor_settings
view is deprecated and removed. To retrieve the configuration parameters, usepg_settings
. - The following columns are removed for PostgreSQL 12 and earlier:
wal_records
,wal_fpi
,wal_bytes
- Changes to
pg_stat_monitor
view for all PostgreSQL versions:
Data type changes | Columns added | Columns renamed | Columns removed |
---|---|---|---|
userid - oid queryid - bigint top_queryid - bigint bucket_start_time - timestamp with time zone |
dbid user |
rows_retrieved to rows plans_calls to calls |
state state_code |
New Features¶
- PG-310 - Add an indication of buckets as “Done”
- PG-400: Add timezone to
bucket_start_time
in thepg_stat_monitor
view - PG-435, PG-543 - Make
pg_stat_monitor
view compatible with that of pg_stat_statements for PostgreSQL 15. - PG-545 - Generate database/scheme/user-agnostic query ID to add visibility into query performance behavior
- PG-586 - Capture CPU and user timing for utility statements
Improvements¶
- PG-354 - Remove obsolete
pg_stat_monitor_settings
view - PG-312, PG-313 - Include queries that take less and more than histogram min/max time to histogram
- PG-321 - Remove redundant
cmd_type
/cmd_type_text
columns inpg_stat_monitor
view. - PG-373 - Remove unused
wal
fields for PG12 and below - PG-488 - Improve the storage mechanism of buckets and query texts using dynamic shared memory for PostgreSQL 15 and onward
- PG-606: Add the ability to enable/disable
pgsm_query_id
calculation via a GUC parameter - PG-607: Allow histogram to track queries in sub-ms time brackets
Bugs Fixed¶
- PG-306 - Fixed datatype for
bucket_start_time
by changing it to timestamp - PG-320 - Removed state/state_code fields from pg_stat_monitor_view.
- PG-518 - Hide internal functions for non-superusers in pg_stat_monitor API.
- PG-552 - Remove unnecessary columns from PostgreSQL 11 and 12 views
- PG-576 - Fixed the issue with incorrect query length in case of a normalized query when the query length exceeds the
PGSM_QUERY_MAX_LEN
value - PG-579 - Added a check if
pg_stat_monitor
is loaded inshared_preload_libraries
before querying the view to prevent the server from crashing - PG-583, PG-597 - Removed rounding off for floating point values
- PG-588 - Fixed missing normalization for some queries
- PG-582 - Fixed rounding for
blk_read_time
andblk_write_time
columns
Get expert help¶
If you need assistance, visit the community forum for comprehensive and free database knowledge, or contact our Percona Database Experts for professional support and services.