Can you configure frigate automatically delete archives when disk usage reaches a certain threshold?

[xrd]

Hi there,

I'm reviewing this:

https://docs.frigate.video/configuration/record#common-recording-configurations

I'm running frigate inside a VM with 100GB disk and today it is at 92%.

My goal is to keep detections and archives as much as possible, but not have to worry it will all fall down if the disk hits 100%.

I'm happy to do use an external technique if this is not something frigate supports natively using a cron job or something.

Is there a way to troubleshoot the configuation so I don't run out of disk? Is there a way to say "retain for 3 days UNLESS you hit 95% and then delete this or that."

I have this configuration right now and worry it is going to crash when the disk gets to 100%

...
record:
  enabled: true
  retain:
    days: 7
    mode: all

...
snapshots:
  enabled: true
  clean_copy: true
  retain:
    default: 60
    objects:
      person: 60
      car: 60
      dog: 60
      cat: 60
...

[inkeep-ai-assistant[bot]]

Frigate does have built-in emergency storage cleanup to prevent disk space issues(1). When there is less than an hour of storage remaining, Frigate automatically deletes the oldest 2 hours of recordings(1). This provides automatic protection against running out of disk space.

However, Frigate doesn't currently support configurable retention by disk usage percentage or size thresholds. The retention system is based on time periods (days) only(1)(2). There is an open feature request for configurable retention by size, but this hasn't been implemented yet(3).

For your current configuration with 7 days of continuous recording (mode: all), you're storing all video segments regardless of activity(1). To reduce storage usage while maintaining protection, consider switching to mode: motion which only keeps segments with detected motion(1):

record:
  enabled: true
  retain:
    days: 7
    mode: motion

(1)

You could also implement a more conservative approach by setting continuous recording to fewer days and using longer retention for alerts and detections(1):

record:
  enabled: true
  retain:
    days: 3
    mode: motion
  alerts:
    retain:
      days: 30
      mode: motion
  detections:
    retain:
      days: 30
      mode: motion

(1)

The emergency cleanup should prevent crashes when disk space is critically low, but it's still better to configure retention periods that fit within your available storage capacity(1).

📚 Sources:

1. Recording
2. Full Reference Config
3. Feature Request - Configurable Retention By Size

---

Have another question? Just tag @inkeep.