What shipped.

Business plans can now run a probe inside their own network. Install it with one command, and it shows up as a thirteenth region on every monitor you point at it.

Private probes run the same checks as the public ones, including screenshots and traceroute, and their results are drawn on status pages in the same bars. They are the answer to the question we were asked most last quarter: can you monitor the thing that is not on the internet?

When several monitors fail within the same ninety seconds and share an upstream, Carbon opens one incident and lists the others inside it. Only the first pages anyone.

• Merging is on by default and can be turned off per monitor group.
• The merged incident resolves when the last monitor in it recovers.
• The post-incident timeline shows every monitor's own recovery time.

Right-click any line in the live tail and choose Add drop rule. Lines matching it are discarded at the edge before ingestion, so they never reach storage and never reach the bill.

Rules match on any field or on a regular expression, and every rule shows how many lines it dropped in the last day, so you can see what it is saving you.

A timeout used to be a single line that said timed out after 30 s. It is now a full traceroute and a ten-cycle MTR report from the probe that failed, attached to the incident.

The report is run at the moment of failure, not afterwards, so it captures the route that was actually broken.

Status pages accept a stylesheet now, in addition to your logo and colours. Change the type, hide the footer, or rebuild the whole thing. Custom domains and CSS are on the free plan too.

Monitors, status pages, on-call schedules, and escalation policies can be declared in Terraform. Import an existing account with terraform import and the provider writes the resources for you.

The provider is open source and lives alongside the REST API it wraps.