Cluster and replication

Topology

database file: main
primaryNode: node-1
replicaNodes: [node-2]
pool: 0

The pool field follows the useful SiriDB concept of grouping servers, but is currently metadata only. AsaDB does not yet distribute one SQL query across multiple files.

Add nodes

On node 1:

curl -X POST https://node-1.example/api/v1/cluster/nodes \
  -H "Authorization: Bearer $ADMIN_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"id":"node-2","url":"https://node-2.example"}'

Register the same physical ID on the replica with primaryNode=node-1 and replicaNodes=["node-2"].

Trigger replication

curl -X POST https://node-1.example/api/v1/replication/main \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"logicalDatabase":"app","targetNode":"node-2"}'

Poll the returned job ID.

Failure behavior

• A primary outage does not automatically promote a replica.
• Operators must verify the latest snapshot and explicitly edit placement.
• A failed snapshot does not mark the replica ready.
• The target restore uses the normal verified `.asb` path.
• Transactions are local to one primary file.

Scheduled snapshots

Set replicationLogicalDatabase on a primary database-file record and set ASADB_REPLICATION_INTERVAL_SECONDS to a positive interval. The maintenance service schedules one snapshot job per configured replica when no equivalent job is already queued or running.

Set the interval to 0 to keep replication manual-only.