at.happyproto.fleet.getRetention

Survival by first-seen cohort. Manual-mode instances are excluded: they have no reporting cadence to miss, so their silence is not churn. Admin only.

Repository @happyproto.at did:plc:mo7vq4jbioxwafy5zkcc3eux namespace authority

Details · query

Kind	query
Authority	fleet.happyproto.at
Indexed	22d ago · 2026-10-19

Compare Edit in builder → Fork →

Query

query · GET

Output

application/json

cohorts

requiredarray

itemsref

Reference

#cohort

have

integer

History available, when accruing.

need

integer

History required, when accruing.

status

requiredstring

Why a figure is absent. Branch on this, never on field presence.

knownValues: ok | accruing | withheld

unit

string

knownValues: days | snapshots

Errors

AuthRequired

This endpoint requires an authenticated caller.

Forbidden

The caller is not a HappyProto admin.

Upstream

The telemetry store could not be queried.

#cohort

object

Properties

cohort

requiredstring

Month the cohort was first seen.

eligible30

integer

Instances old enough to have survived 30 days. THIS is the denominator, not `started` — a cohort ten days old has not failed to survive thirty.

eligible60

integer

eligible90

integer

started

requiredinteger

survived30

integer

survived60

integer

survived90

integer

Similar lexicons

semantic match

Lexicons whose schemas describe something close to this one. Check them before publishing a new one that overlaps.

at.happyproto.fleet.getSignals
query

Fleet-wide failure signals, derived from each instance's most recent snapshot. Admin only.
at.happyproto.fleet.getBenchmarks
query

Where one instance sits in its size cohort. Withheld below a cohort of 10: a percentile over fewer is not an aggregate, it is individual data with a statistic drawn over it. Admin only.
at.happyproto.fleet.getAdoption
query

Per-subsystem feature adoption, ever-used against active now. Admin only.
at.happyproto.fleet.getVersions
query

HappyView version distribution and upgrade lag. Admin only.
at.happyproto.fleet.getResets
query

Counter resets — the negative deltas instance_daily_deltas discards. A reset is a restored backup, a wiped volume, a re-provision or a wrong clock. Admin only.