Client Closed Request on the container's activity endpoint

Activity operations have no explicit timeout, and after pg_restore completes, the database likely enters a state where connections are hanging. It's likely the HTTP request has a timeout. The client returns (499) before the psql command actually fails.

172.17.0.1 - - [12/Nov/2025:00:30:32 +0000] "GET /api/instance/retrieval HTTP/1.1" 499 0 "-" "Client/1.2.21" "-"
172.17.0.1 - - [12/Nov/2025:00:30:52 +0000] "GET /api/instance/retrieval HTTP/1.1" 499 0 "-" "Client/1.2.21" "-"