pgCluu 5.0 adds PostgreSQL I/O stats, cgroup limits and kernel pressure reports

pgCluu 5.0, a Perl-based PostgreSQL cluster auditing tool, has been released with expanded system-level visibility. The update adds collection of I/O statistics via pg_stat_io for PostgreSQL 16+, reports on replication slots and pg_hba_file_rules, and tracks postmaster process limits including cgroup v2 memory, CPU and I/O quotas. It also introduces kernel pressure stall information (PSI) reporting and memory events for both the host and PostgreSQL's cgroup.

What's new in pgCluu 5.0

• I/O statistics: Collects from pg_stat_io (PostgreSQL 16+), with reports for throughput per backend type, shared buffers, WAL I/O, temporary relations I/O, and async I/O (PostgreSQL 18+)
• Postmaster limits: Tracks oom_score_adj, file/process/memory limits, and cgroup v2 constraints (memory.max, cpu.max, io.max, pids.max) plus v1 quotas
• Kernel parameters: Expanded collection to include vm.dirty_*, NUMA settings, transparent hugepage shmem config, block device tunables, and CPU frequency governor data
• Pressure stall information: Reports CPU/memory/I/O pressure percentages for both host and PostgreSQL cgroup via /proc/pressure (kernel 4.20+)
• Memory events: Tracks cgroup memory.events (high, max, oom, oom_kill) and /proc/vmstat counters including allocstall, compaction, and page cache refaults

Why it matters

The expanded kernel and cgroup visibility helps database administrators diagnose resource contention and OOM-related stability issues in containerized or tightly regulated production environments.