Monitor a fleet

Track collector instances, their status, and the configuration they run.

This page describes how to monitor a server fleet from its detail page. The Client Instances tab lists every instance that has connected, with its status, the configuration version it runs, its token sync state, its collector version, and when it was last seen. The page refreshes automatically, and you can refresh it manually. A stats strip at the top summarizes the fleet: total instances, how many are online, version drift, and when the token expires.

This page covers server fleets. Kubernetes fleets report through a bridge and use a different status model. See Kubernetes fleets.

Instance status

Telflo derives each instance's status from the version it has been served, the version it has been assigned, and what the collector reports.

StatusMeaning
In syncThe collector applied the configuration it was assigned.
Pending pickupA new version is assigned, but the collector has not received it yet.
ApplyingThe collector received the version and is applying it, or has not reported a result yet.
Apply failedThe collector rejected the configuration. It needs attention.
StaleNo recent heartbeat. The collector has gone quiet.

A colored dot accompanies each status: green for in sync, amber for pending pickup and applying, red for apply failed, and gray for stale.

Instance status lifecycleROLLOUT STATESPending pickupassigned, not yet receivedchecks inApplyingreceived, applying nowappliedIn syncruns the assigned versionrejectedApply failedfix, publish, roll out againStaleno heartbeat for 15 mingoes quiet
How an instance converges on an assigned version. A rejected configuration reports Apply failed, and a collector with no recent heartbeat is Stale.

Online and stale

Instances report on a heartbeat.

  • Online: confirmed within the last 2 minutes.
  • Stale: no contact for more than 15 minutes. An instance that already reported In sync or Apply failed keeps that status; Stale applies to instances that have not yet confirmed their assigned version.

An apply failure stays visible even if the collector then goes quiet, so the failure is not masked by a missing heartbeat. Inspect the configuration and the collector logs, fix the issue, and roll out a corrected version.

Manage instances

From the fleet detail page you can:

  • Rename an instance.
  • Migrate instances to a different published configuration version with Migrate Instances.
  • Delete a decommissioned instance.

The detail page also has Client Tokens and Deployment Tokens tabs for the fleet's tokens. See Tokens & security.

Next steps

Last updated on

On this page