NOREPLICAS error

The primary accepts writes only when enough replicas are connected, and there aren't enough.

A primary returns this error when you send a write, but fewer replicas are connected than its min-replicas-to-write setting requires. This setting limits how much data you can lose if the primary fails. Reads still work. Writes start working again when enough replicas are connected.

A replica only counts if it's up to date. Each replica reports to its primary every second. If a replica hasn't reported for more than min-replicas-max-lag seconds, the primary doesn't count it.

NOREPLICAS Not enough good replicas to write.

How to fix it

1. On the primary, run INFO replication. connected_slaves shows how many replicas are connected. Each slave0, slave1, and so on line shows a replica's address and its lag in seconds.
2. Fix the replicas that are missing or lagging. Check that they're running and can reach the primary. On a replica, INFO replication shows the state of its link to the primary.
3. If writes must work with fewer replicas, lower min-replicas-to-write. Set it to 0 to turn the check off.

See Allow writes only with N attached replicas.