One pane for every Proxmox Backup Server you run
Syntera is our answer to scattered PBS instances: enroll each server once, and get centralized telemetry, datastore timelines, alert configuration, and a React interface backed by TimescaleDB. It is architected for AI agents from the start — the same system that watches our own backup fleet.
WHAT IT DOES
Syntera collects telemetry from enrolled Proxmox Backup Server instances into TimescaleDB and renders it as real-time dashboards: datastore timelines, task history, deduplication ratios, and verification status across the whole fleet. Alerting is configurable per-instance, and the three-tier role model (guest help-only, read-only user, full admin) maps onto how teams actually delegate backup oversight.
Authentication supports a local realm and LDAP/AD integration, so it drops into an existing directory rather than inventing another account silo. The MCP server exposes the fleet to AI agents — autonomous monitoring, decision-making, and a clean handoff surface between human engineers and agent workflows.
SEE IT RUNNING
Syntera monitors our own production backup fleet. We don't expose the instance publicly — but we will happily walk you through it live: dashboards, datastore timelines, alert configuration, and the agent integration against real backup data. Ask for a guided walkthrough — a running system beats a slide deck.
WHY IT EXISTS
We run Proxmox fleets for ourselves and for clients. "Backups exist" is easy; "backups are verifiably recoverable across twelve instances" is a different discipline — restore-test cadence, per-datastore trends, and alerting that a human actually reads. Syntera is the control plane for that discipline, and the same engineering pattern we bring to observability engagements: measure, alert on signal, and keep the data on infrastructure you own.
Backup fleet outgrowing per-server habits?
Whether it's Syntera or a management plane built for your exact stack, the pattern is the same: centralize telemetry, alert on signal, automate the routine.