---
title: Metrics
description: "Metrics of each server and database service on the Monitoring tab, metrics retention, PromQL queries, and federation into your own Prometheus."
sidebar:
  icon: activity
search:
  keywords: ["metrics", "Prometheus", "Grafana", "PromQL"]
---

The console collects metrics from each database server and keeps them for 30
days by default. For alerts, see [Alerts](/docs/v2026.11.6-beta/alerts).

## The Monitoring tab

The **Monitoring** tab of a service has seven panels:

| Panel | Shows |
| --- | --- |
| CPU | CPU use (%) |
| Memory | Memory use (%). Hover to see the bytes. |
| Storage | Database size and disk used |
| Transactions per second | Commits plus rollbacks |
| Disk IOPS | Read and write operations per second |
| Disk bandwidth | Read and write bytes per second |
| Connections | Total and active. Hover to see `max_connections`. |

Each panel has one chart for each PostgreSQL server, primaries first. A
service with a replica has a node filter: **All nodes**, **Primary**, or one
server.

![The Monitoring tab of a service, with the CPU, memory and storage charts](/docs/screenshots/v2026.11.6-beta/monitoring/service-dashboard.light.webp)

- The time range goes up to **Last 30 days**.
- Red shading marks a critical [alert](/docs/v2026.11.6-beta/alerts) of the service, amber a
  warning. Hover to see the alerts.
- CPU and Memory show percent of the full server. When the
  [license](/docs/v2026.11.6-beta/licensing) limits TimescaleDB, a dashed **License limit** line
  shows the limit (for example 25% for a 4 vCPU license on a 16 vCPU server).

## Monitoring settings

Open **Administration → Monitoring** (admin only).

![The Monitoring page of Administration, with Keep metrics for and Metrics mirroring](/docs/screenshots/v2026.11.6-beta/monitoring/monitoring-settings.light.webp)

- **Keep metrics for**: 1 to 3650 days (default 30). The page shows the disk
  space that the metrics use. A shorter time deletes the older metrics. After
  a save, the charts get no data for some seconds.
- **Metrics mirroring**: lets your own Prometheus copy the metrics. See
  [Federate into your own Prometheus](#federate-into-your-own-prometheus).

## PromQL queries

To query the metrics with PromQL, use `POST /api/v1/promql/query` or
`POST /api/v1/promql/query_range` (viewer role). They return Prometheus JSON.
See the [API reference](/docs/v2026.11.6-beta/reference).

## Federate into your own Prometheus

1. **Make a token**

    In **Administration → Monitoring**, turn on **Metrics mirroring** and
    select **Generate token**. Copy the token. The console shows it only
    once.

2. **Scrape the console**

    Add this job to your Prometheus:

    ```yaml
    scrape_configs:
      - job_name: timescale-enterprise
        honor_labels: true
        metrics_path: /metrics/federate
        params:
          "match[]": ['{__name__=~".+"}']
        authorization:
          credentials: <token>
        static_configs:
          - targets: ["<console-host>"]
    ```

`/metrics/federate` returns 404 while mirroring is off or there is no token.
**Rotate token** makes a new token. The old token stops at once.

The export also has these metrics of the console:

| Metric | Value |
| --- | --- |
| `tse_console_up` | Always `1`. No series: the console does not answer. |
| `tse_backup_needs_attention{service}` | `1` when the backups of the service need attention |
| `tse_server_enrolled{server}` | `1` when enrollment is complete, `0` while pending |
| `tse_license_expiry_timestamp_seconds{licence_id, licence}` | Expiry time of each installed license with an end date |
| `tse_license_expired{licence_id, licence}` | `1` for an expired license that a service uses, or when no active license is installed |
| `tse_console_clock_offset_seconds{source}` | Seconds to add to the console clock (`source` is `ntp` or `servers`). See [Console clock](/docs/v2026.11.6-beta/configuration#console-clock). |
| `tse_console_clock_offset_known` | `1` when the console knows its clock offset |
| `tse_health_read_duration_seconds` | Histogram of the time of one service health read |
| `tse_health_reads_due` | Services with a due health read that did not start |
| `tse_health_oldest_check_age_seconds` | Age of the oldest last health check of a running service |
| `tse_console_backup_last_success_timestamp_seconds{target,schedule}` | Time of the newest console backup. Manual backups have no `schedule`. |
| `tse_console_backup_last_run_failed{target,schedule}` | `1` when the last console backup failed |
| `tse_console_backup_last_run_duration_seconds{target,schedule}` | Duration of the last console backup |
| `tse_console_backup_archives{target}` | Console backups on the target, also those of other consoles. Listed every 10 minutes. |
| `tse_console_backup_archive_bytes{target}` | Size of the console backups on the target |
| `tse_console_backup_key_confirmed` | `1` when an admin confirmed the recovery key. Without it, no console backup runs. |
| `tse_console_backup_schedules_enabled` | Console backup schedules that are on |
| `tse_console_backup_schedule_stale_after_seconds{target,schedule,cron}` | Backup age at which `ConsoleBackupStale` fires |

Starter dashboards for your own Grafana:
[postgres-exporter](/docs/grafana-dashboards/postgres-exporter.json),
[node-exporter](/docs/grafana-dashboards/node-exporter.json),
[patroni](/docs/grafana-dashboards/patroni.json),
[haproxy](/docs/grafana-dashboards/haproxy.json),
[pgbouncer-exporter](/docs/grafana-dashboards/pgbouncer-exporter.json), and
[timescaledb](/docs/grafana-dashboards/timescaledb.json).
