Dashboards

Two read-only views over everything the account already has.

Financial

What the fleet costs, dearest first.

The figure is an estimate: list price times what each project holds. Providers do not expose a billing endpoint KamalCloud can read, so this is arithmetic over their public price lists, not your invoice. It will not match to the cent, and it is not meant to.

What it breaks down by:

• Kind – servers, backups, primary IPs, floating IPs, volumes, load balancers, snapshots
• Location
• Machine type

It also shows:

• Idle spend – money going out for something switched off or attached to nothing. A powered-down machine still costs; a volume attached to nothing definitely does.
• Traffic – usage past each machine’s included allowance, projected linearly from the month so far, which is the best a month-to-date counter supports.

Self-hosted machines are absent on purpose. KamalCloud has no price for hardware you own, and an invented figure inside a total is worse than a machine missing from one.

Server rows come from KamalCloud’s own records, synced every few minutes. Everything else is read from the provider when the page is built. The whole thing refreshes on a six-hourly schedule – which is right for data that changes at the provider rather than here – and Refresh is there for when it just did.

Each day’s total is written to a snapshot, so a past day can be explained later rather than recomputed from prices that have since changed.

Analytics

How the fleet is behaving.

Deployments by day, split by outcome. Only releasing deployments are counted – maintenance flips and lock releases are not deployments in any sense a chart should count. Days with nothing in them are filled in, so the axis has no gaps.

Deploy runtime, averaged per day over successful releases. A day with no release breaks the line rather than dropping it to zero.

Backups by day, counting both accessory dumps and volume archives – a chart counting only dumps would tell an account backing up nothing but volumes that it was backing up nothing at all.

Spend by day, from the snapshots. It starts where the recording did rather than pretending the spend was zero before that.

Machine metrics – CPU, traffic, disk – read from records KamalCloud keeps rather than from the provider on every page load. Pick a server and a range.