Release LoadGen Api, Appliance and MCP Server: 1.0.0.11932

Updated June 10, 2026 13:41

LoadGen General

SessionSight ingestion recovery. Fixed SessionSight appearing to lose data after a container restart: when an ingestion worker failed to start, incoming events could be dropped while the worker was down. Failed or stuck workers now recover automatically, and ingestion resumes the moment a worker comes back. Worker start-up timeouts now log detailed diagnostics so the root cause of a failed start can be pinpointed immediately.
SessionSight disk spooling for unavailable workers. SessionSight events that arrive while the ingestion worker is briefly down (cold start, restart) are no longer rejected with a 503 — the API now spools them to disk and answers 202 Accepted with the response header X-LoadGen-SessionSight-Spooled: true, then replays them automatically in original capture order shortly after the worker returns. The spool is bounded and lives on the API's data volume so spooled events survive a restart of the API container, and data that repeatedly fails to replay is moved aside for investigation instead of blocking newer events. The image ingest endpoint is out of scope and keeps its existing behavior.

[#8639] The Setup PostgreSQL action on Infrastructure → Appliance → Services previously returned "PostgreSQL setup is not yet implemented on this appliance. The provisioning feature ships in a future update." — provisioning is now fully implemented: POST /appliance/postgres/setup creates the Service Levels database (loadgenservicelevels), the loadgen schema, and two dedicated roles (loadgenadmin for schema management, loadgen_app for data access, with default privileges so tables created later are automatically accessible) with strong generated passwords that are echoed once and stored encrypted on the appliance. Setup is idempotent — re-running it returns the existing configuration instead of rotating passwords — and PUT /appliance/postgres/credentials rotates either role's password on demand. GET /appliance/postgres/status now reports the database, schema, and role names, and GET /appliance/postgres/discovery enumerates databases, roles, and schemas even when the saved configuration is missing (e.g. after a data-volume restore), with guidance to rotate the role passwords back to a known-good state. Failures surface PostgreSQL error codes (e.g. 42710 name already exists, 28P01 invalid authorization) rather than generic messages.