SnapStream

Feed

Coming soon

feed.json is SnapStream's public status endpoint. During a coordinated cluster restart, it publishes a snapshot of who is participating, how far each one has gotten, and how the group's stake adds up against the threshold needed to restart.

What each snapshot carries

Four things, at a glance:

  • Participant state: each participant moves through a small set of named states, from offline through acknowledging the restart plan to voting again. The feed reports the current state, not a raw progress number.
  • Stake weighting: restart progress is measured in stake, not in a headcount of validators, and every percentage field names what it's a percentage of. That basis can change between one coordination event and the next, so the feed states it each time rather than assuming readers remember.
  • Client identity: a participant's validator client and version are recorded as either detected automatically or declared by the operator, and the feed marks which of the two it is.
  • Rehearsal flag: test runs built from fixture data carry an explicit rehearsal flag, so a dashboard or script can tell a drill from a real restart without guessing.

Not live, and never load-bearing

Publishing the feed is a separate, operator-armed switch, not something that turns on by itself once the underlying code is ready. It goes live with the first public coordination event.

The feed is a status window, not a control plane. If it is unreachable or wrong, no validator's restart depends on it, and voting continues on its own.

← Back home