ImmichI
Immich•4w ago
Valle

Server API seems to freeze while Machine Learning unhealthy

Every now and then the server API freezes for a few minutes for no apparent reason. This is especially visible when switching to admin pages.

I inspected the logs while the freeze was happening and it precisely coincided with Machine Learning becoming unhealthy and then healthy, again.
CPU and network were idle during that time. docker-compose and env are default.

I haven't found any exact report of this issue, yet.

Version: 2.3.1 (docker)

The API was unresponsive for the 3 minutes Machine Learning was unhealthy:
immich_server            | [Nest] 23  - 11/23/2025, 2:34:14 PM     LOG [Api:MachineLearningRepository] Machine learning server became unhealthy (http://immich-machine-learning:3003).
immich_server            | [Nest] 23  - 11/23/2025, 2:34:27 PM     LOG [Api:WebsocketRepository] Websocket Disconnect: A-fiItbFiCVdEsq0AAAD
immich_redis             | 1:M 23 Nov 2025 13:34:55.086 * 100 changes in 300 seconds. Saving...
immich_redis             | 1:M 23 Nov 2025 13:34:55.087 * Background saving started by pid 77
immich_redis             | 77:C 23 Nov 2025 13:34:55.098 * DB saved on disk
immich_redis             | 77:C 23 Nov 2025 13:34:55.099 * Fork CoW for RDB: current 0 MB, peak 0 MB, average 0 MB
immich_redis             | 1:M 23 Nov 2025 13:34:55.188 * Background saving terminated with success
immich_server            | [Nest] 23  - 11/23/2025, 2:37:14 PM     LOG [Api:MachineLearningRepository] Machine learning server became healthy (http://immich-machine-learning:3003).
immich_server            | [Nest] 23  - 11/23/2025, 2:37:15 PM     LOG [Api:WebsocketRepository] Websocket Connect:    F_sk2KnqGc-P4sQVAAAF
Was this page helpful?