MONITORING

Continuous watch over every resource.

Health checks, metrics, alerts and live log streaming across every resource Talos manages, with a dashboard shaped for what you run rather than one generic view for everything.

Install itSee the install guide
SSE·Health checks·Per-type dashboards·Email / webhook alerts

THE MANUAL WAY

Finding out from the customer.

THE SILENT FAILURE

A nightly job has failed for eleven days. It exits non-zero into a log nobody reads, and the first sign of trouble is a customer email.

THE WRONG DASHBOARD

Every project shows the same four charts, none of which are the numbers that matter for a database, a cluster or an ERP site.

THE ALERT NOBODY TRUSTS

Alerts fire so often that the channel is muted. The one that mattered was in there somewhere.

WHAT TALOS DOES

Signals worth reacting to.

Per-type dashboards

A database project shows backup age and verification status; a cluster shows node pressure. Each kind of project has its own view rather than a shared default.

Health checks

Reachability, resource pressure and service state, polled on a schedule and recorded so trends are visible.

Live log streaming

Follow any running job or container as it happens. The stream is the same thing that lands in the record.

Alerts that name the resource

Alerts are scoped to a resource and an organization, so they reach the people responsible for it and nobody else.

Job failure attribution

A failed job points at the change, the parameters and the person, rather than at a stack trace alone.

Fleet overview

One page for the whole estate, with the exceptions surfaced instead of buried in a list of green.

talos.internal/monitoring
Servers
Cloud
Kubernetes
Docker
Databases
Frappe
Pipelines
Monitoring
MLOps
Proxy
DNS & SSL
MCP
Monitoring86 resources · 2 alerts
acme.erpHEALTHY
cpu 18% · mem 41% · disk 62%
prod-eu · nodesHEALTHY
12 ready · 0 pressure
legacy-crm backupALERT
unverified for 6 days
db-01 · diskALERT
88% used · +4%/week
! backup verification failed · legacy-crm
→ last verified 2026-07-28
→ notified 2 people with db.backup access
! db-01 disk 88% · projected full in 21d
✓ 84 resources healthy

HOW IT RUNS

Every action is a tracked job.

Requestyou, API or schedule
Queueapproval if required
Job runnerbackground worker
ExecutionSSH · Ansible · Terraform · API
Your infrastructurelogs stream back live

WHAT MAKES THIS DIFFERENT

Every alert reaches the right organization, and only that one.

An alert raised for one client’s infrastructure is visible only to that client. Alerts are scoped to the organization that owns the resource, and the scoping is enforced in the data path rather than by a filter each view has to remember.

Alerts route by who holds permission on the resource, not by a global notification list.
Every kind of project has a dashboard built for its own metrics.
Clusters attached by kubeconfig are monitored like any other cluster.
alert routing · legacy-crm
✓
resource scoped
organization: northwind-ops
OK
✓
recipients resolved
2 people hold db.backup on this resource
SENT
✕
other organizations
not notified · not visible
SCOPED
✓
recorded
alert and its resolution in the ledger
DONE

RELATED CAPABILITIES

Servers

The hosts most of these signals come from.

explore →

Databases

Backup age, drift and verification failures.

explore →

AI Engine

Turn a failing job into a reviewable proposed fix.

explore →

Start by monitoring one server.

Connect a server and the health checks start immediately. No agent, no exporter to configure.

Install in one commandSee every capability →