Problem / Motivation
All staged and compiled protos live in process memory. When the backend disconnects or Postgres restarts everything is lost. Users have to re-stage and recompile every session. In a cluster setup protos would need to be manually configured on each node.
Proposed Solution
Store protos in an internal Postgres table (e.g. pg_grpc.proto_files). This survives restarts, works with pg_dump/backup and replicates to standby nodes via WAL. Load into the in-memory registry on backend startup.
Problem / Motivation
All staged and compiled protos live in process memory. When the backend disconnects or Postgres restarts everything is lost. Users have to re-stage and recompile every session. In a cluster setup protos would need to be manually configured on each node.
Proposed Solution
Store protos in an internal Postgres table (e.g.
pg_grpc.proto_files). This survives restarts, works with pg_dump/backup and replicates to standby nodes via WAL. Load into the in-memory registry on backend startup.