GET /mdp/meta

This endpoint is an out-of-band probe for transport and deployment discovery.

It lets another local process check whether a port is serving MDP before trying to open a client transport. The current server implementation uses it for peer discovery, cluster control bootstrap, and upstream proxy setup.

Request

GET /mdp/meta
Accept: application/json

Response

{
  "protocol": "mdp",
  "protocolVersion": "2.0.0",
  "supportedProtocolRanges": ["^2.0.0"],
  "serverId": "127.0.0.1:47372",
  "endpoints": {
    "ws": "ws://127.0.0.1:47372",
    "httpLoop": "http://127.0.0.1:47372/mdp/http-loop",
    "auth": "http://127.0.0.1:47372/mdp/auth",
    "meta": "http://127.0.0.1:47372/mdp/meta",
    "cluster": "ws://127.0.0.1:47372/mdp/cluster"
  },
  "features": {
    "upstreamProxy": true,
    "clusterControl": true
  },
  "cluster": {
    "id": "127.0.0.1:47372",
    "membershipMode": "dynamic",
    "membershipFingerprint": "dynamic",
    "role": "leader",
    "term": 3,
    "leaderId": "127.0.0.1:47372",
    "leaderUrl": "ws://127.0.0.1:47372",
    "leaseDurationMs": 4000,
    "knownMemberCount": 3,
    "reachableMemberCount": 3,
    "quorumSize": 2,
    "hasQuorum": true
  }
}

Notes

• This is not an MDP wire message.
• It does not require a websocket or HTTP loop session.
• Clients that already know the correct serverUrl do not need to call it.
• protocolVersion is the server's exact protocol semver.
• supportedProtocolRanges contains semver ranges accepted for server-to-server proxying. Exact versions are still valid ranges.
• A proxy-capable server should verify that its required protocol version satisfies one of those ranges before mirroring local clients into an upstream hub.
• endpoints.cluster is the server-to-server control websocket used for heartbeats, leader resignation, and elections.
• cluster.id is the logical cluster identity. Discovery and server-to-server control traffic should only join peers that advertise the expected cluster id.
• cluster.membershipMode tells you whether this node is running with discovery-driven membership (dynamic) or one explicit configured member set (static).
• cluster.membershipFingerprint is the compatibility key for that membership mode. Static peers in one cluster should agree on it exactly.
• serverId must be unique within one logical cluster. A duplicate server id on another endpoint is treated as a cluster configuration error.
• cluster.role, cluster.term, and cluster.leaderId describe the server's current HA view.
• cluster.leaseDurationMs is the follower lease window for primary heartbeats.
• cluster.knownMemberCount is the membership size known to this process. It may come from sticky discovery state or from an explicit static member list.
• cluster.reachableMemberCount is this node's current local view of how many members, including itself, are reachable within the lease window.
• cluster.quorumSize is the current majority threshold derived from the known member set.
• cluster.hasQuorum tells you whether this node currently sees enough reachable members to satisfy quorum.
• The current implementation keeps cluster membership in memory for the lifetime of the process; quorum does not automatically shrink just because discovery temporarily stops seeing a peer.