The pgCluu team has announced the release of v5.0, a major update to the Perl-based PostgreSQL cluster audit tool. pgCluu works in two stages: a collector gathers statistics from the server via psql and sar, and a report builder turns that data into HTML pages and charts. This release adds several new reports, broadens what the collector captures, and fixes issues reported since the previous version.
- Added a report on replication slots, populated from
pg_replication_slots. - Added collection of
pg_hba_file_rules, bringing host-based authentication rules into the audit output. - Wait event reporting is now limited to events representing more than 0.1 percent of all observed wait events, keeping reports focused on meaningful bottlenecks.
- Added collection of I/O statistics from
pg_stat_ioon PostgreSQL 16 and later. - New I/O reports cover: throughput per backend type; shared buffers, writers and evictions; buffer hit ratio and ring buffers; I/O latency and size; WAL I/O; temporary relations I/O; and asynchronous I/O on PostgreSQL 18 and later.
- More kernel tuning parameters are collected, and
shmmax/shmallwere removed. The parametersvm.dirty_expire_centisecsandvm.dirty_writeback_centisecswere previously not matched by thesysctlfilter; this is corrected, along with othervm.dirty_*parameters.
For developers and DBAs running PostgreSQL in production, the shift toward pg_stat_io is the most consequential change. Before PostgreSQL 16, I/O behaviour had to be inferred indirectly; pgCluu v5.0 now surfaces backend-type throughput, WAL and temporary-relation I/O, and buffer eviction pressure in one place. Combined with the replication slot and authentication rule reports, this makes pgCluu a more complete audit of cluster health, useful both for periodic capacity reviews and for diagnosing slow queries tied to I/O waits.
Source: https://www.postgresql.org/about/news/pgcluu-v50-has-been-released-3417/