An autonomous recovery agent that extends CloudNativePG with self-healing
CloudNativePG already runs Postgres on Kubernetes with solid built-in automation. Veritas sits beside it as an autonomous recovery layer — watching the cluster, detecting incidents before they cascade, and reconciling state without paging a human. Same operators, same reconcile loop, fewer 3 a.m. pages.
What you have, and what sits on top
The left column is the platform you already trust; the right column is the autonomous loop Veritas closes around it.
Mature operators manage the lifecycle of PostgreSQL instances on Kubernetes — failover, scaling, point-in-time recovery, fencing.
Built-in automation handles the well-known paths: replica promotion, PVC resizing, scheduled maintenance.
Alerts fire when something goes off the happy path; an SRE then triages, decides, and remediates.
Strong foundation for running Postgres at scale, but the recovery loop still relies on a human being awake.
Agents on every node stream replication lag, WAL pressure, and connection health into a continuous detection engine that flags anomalies under thirty seconds.
The recovery engine generates a fix plan, evaluates its risk, and executes it directly against the cluster — failing over, rebalancing, restarting — without waiting for an alert to land in a channel.
A verification step confirms the fix took hold and reconciles state against the declared intent; the full incident is logged with annotations for your team after the fact.
Sits beside CloudNativePG rather than replacing it — same operator surface, same backups, plus a closed-loop recovery loop running 24/7.
Want a 20-minute walkthrough on the floor?
Drop us a line — we’ll find a slot between talks.
November 2026 · dates pending
Tokyo · venue to be announced
Cluster reliability with autonomous recovery (placeholder)