Usage over time

Who is using the cluster, and when

{{end}} {{/* One row of controls above the charts, which is where a reader looks for them. Links rather than a script: the choice is in the URL, so a view worth coming back to can be bookmarked or pasted to somebody else. */}}

{{template "tschart" dict "t" .Requests "title" "Requests" "note" "Requests served in each slice, counted where the key belongs now rather than where it was at the time."}} {{template "tschart" dict "t" .Tokens "title" "Tokens" "note" "Prompt and completion together, which is the figure a person recognises as what they consumed."}}

{{if .Priced}} {{template "tschart" dict "t" .Spend "title" "Spend" "note" "In whole units, the same ones a budget is written in."}} {{else}} {{/* Said rather than drawn. A spend chart flat along the bottom reads as a quiet month, and this is not that. */}}

Spend

No model has a price, so every request costs zero and this chart would be a flat line however busy the cluster is. Set prices in the pricing block of the server configuration.

Group	Requests	Tokens	Spend	Errors
{{.Name}}	{{compact .Requests}}	{{compact .Tokens}}	{{.Spend}}	{{if .Errors}}{{.Errors}}{{else}}—{{end}}

Nothing recorded in this window.

{{end}} {{/* The charts fold their tail; the table does not. Both are here because they answer different questions, and the one that is easy to get wrong is thinking the four coloured lines are all of it. */}}

The charts draw the four busiest groups by total and sum the rest into “other”, so they always add up to the whole cluster. The table below is every group, in the same order.

A person is not a level of the quota tree — one person holds keys in several projects — so grouping by person crosses the tree rather than naming a level of it. Quotas are written on the other three, on the quotas page.