Fleet flow view¶
The Flow view answers a question Inventory cannot: where is my fleet's telemetry actually going?
It renders a graph with:
- left: agents (or aggregated agent groups), grouped by tenant and group,
- middle: pipelines per signal,
- right: exporter destinations.
Edges thickness represents the number of agents using that path. Hover any edge to see the list of agents on it.
Aggregation toggle¶
The flow view has two aggregation modes:
- By alias — every distinct exporter alias is its own node (
otlp/staging,otlp/prod,prometheusremotewrite/central). Faithful but busy. - By type — exporter aliases are collapsed by type (
otlp,prometheusremotewrite,kafka). Less precise but usually what you want for "where is the data going at the macro level".
Toggle in the toolbar.
Caching¶
The flow graph is expensive to compute, so it is cached per tenant for 30 seconds and invalidated on group / membership / configuration events. You will not see "live, instantaneous" updates here — that is intentional. Click Refresh to force a recompute.
Filtering¶
- Group filter — render only the agents in a specific group.
- Signal filter — show only
traces,metrics,logs, orprofiles. - Exporter type filter — show only
otlp, onlykafka, etc.
Filters compose, the same way they do on the Inventory page.
Use cases¶
- Migration planning — "we want to move all
otlp/stagingtraffic to a new endpoint"; the flow view shows you which agents are affected. - Incident response — "exporter
otlp/centralis down"; flow view highlights every agent feeding it. - Compliance review — "no agent should export to a non-EU endpoint"; visual sanity check before audit.