503 Service Unavailable Response Conditions

Occasionally Swarm may return a 503 Service Unavailable response to the client. The following is a list of known possible causes:

• A SAN (Secondary Access Node) will respond with a 503 if a request is directed to it and there is insufficient memory to complete the request.
• A SAN will respond with a 503 on an indexer listing request without an indexer available.
• On pre-7.0.0 Swarm nodes, poorly formed Indexer (now Swarm Search) requests previously returned a 503 response but as of Swarm v7.0.0 return a 400 Bad Request
• A PAN (Primary Access Node) will 503 if all nodes are in initialization (prior to reading the Persistent Settings Stream). This will happen during the short period on Swarm startup after the Swarm process has started but before the settings stream has been read. If the node was not used as a PAN, then this condition will not occur.
• In Swarm v6.1.0, it was possible to get temporary 503 responses when modifying the Persistent Settings Stream. This was addresses in the v6.1.1 patch release.
• On pre-Swarm v6.1.2 clusters, if writing with replicate-on-write (ROW) while the cluster was under heavy load, a node suddenly disappearing could cause 503 responses. The v6.1.2 patch addressed this issue more gracefully.