YSQL cluster administration features

[rkarthik007]

Jira Link: DB-1393
Issue tracking improvements needed around administration of YSQL.

Design Docs

⬜️ Design Doc for cluster administration

Node level statistics

In this phase, each YSQL server will maintain configuration and statistics about queries executed on that node. These statistics can be queried using SQL statements via ysqlsh or any postgres client driver. However, the statistics across all nodes won't be aggregated. Some stats will also be available on tserver UI page.

Status	Feature	Comments
✅	Setting YSQL configuration per node (pg_hba.conf). Available configuration settings	v2.5
✅	Support for pg_stat_statements	v2.5 (default enabled)
✅	Support PGAudit	v2.5 (default bundled)
✅	Support for slow query logging #3657	v2.5
✅	Support pg_stat_activity	v2.5 (default enabled)

Table level statistics

Status	Feature	Comments
⬜️	Track statistics on table and index accesses, like pg_stat_all_tables
⬜️	Track statistics on usage of user defined functions, like pg_stat_user_functions
⬜️	Database statistics like number of connections, transactions committed / rolled back, rows read and written, similar to pg_stat_database

Cluster wide statistics

In this phase, each YSQL server will collect statistics about data and queries executed on that server and will push these out to a centralized statistics table. This table can then be queried using SQL statements. These stats will also be available on YW platform and master UI pages.

Status	Feature	Comments
⬜️	Add ability to change YSQL configuration for the entire cluster #1140
⬜️	Track SQL statement execution with pg_stat_statements #4802
⬜️	Track statistics on DB, table, index accesses and user defined functions pg_stat_xx

Visibility into transactions and locking

Status	Feature	Comments
⬜️	Viewing locks in the system with pg_locks
⬜️	View statistics such as read / write action counts taken so far in transactions, like pg_stat_xact_all_tables