0.15 rc2 | no space left on device

[FeatherKing]

Ive been having this problem on all the 0.15 builds. Ive slowly increased the cache and SHM but im still having this happen every few days. I am not sure what device this error is referencing as i have space in recording, /tmp/cache, and SHM. The SHM is 86% full but ive never seen it go higher than that. The SHM was recommended when i first started 0.15. I will try bumping that but wanted to post here.

frigate | 2025-02-08 10:08:19.554760758 [2025-02-08 10:08:19] ffmpeg.hagdrivewaycam.record ERROR : [out#0/segment @ 0x55c8a565bec0] Could not write header (incorrect codec parameters ?): No space left on device
frigate | 2025-02-08 10:08:19.554780439 [2025-02-08 10:08:19] watchdog.hagdrivewaycam INFO : Terminating the existing ffmpeg process...
frigate | 2025-02-08 10:08:19.554823838 [2025-02-08 10:08:19] watchdog.hagdrivewaycam INFO : Waiting for ffmpeg to exit gracefully...

docker compose set like this

services:
frigate:
container_name: frigate
privileged: true # this may not be necessary for all setups
restart: unless-stopped
image: ghcr.io/blakeblackshear/frigate:0.15.0-rc2
shm_size: "580mb"
tmpfs:
- /tmp/cache:size=512m

as i am writing this post i see inside the container that cache is now totally full

Filesystem Size Used Avail Use% Mounted on
tank/d1907ddf9fd95a90372c0c8b30c8f9c8d5b55cd80ba79cc6337fb2819d6b6186 16T 2.1G 16T 1% /
tmpfs 64M 0 64M 0% /dev
shm 580M 505M 76M 87% /dev/shm
tank 26T 9.5T 16T 38% /config
tmpfs 512M 512M 0 100% /tmp/cache
/dev/mapper/rl_nas-root 26G 4.5G 21G 18% /usr/share/zoneinfo/Etc/UTC

ill increase that but when i took the above screenshot it was only 18%, so ive had a sudden jump from 18% used to 100%. Ill try to search the logs for when it started

[FeatherKing]

ill just attach the logs here. Seeing a "temporary failure in name resolution"? and then it was all full. weird

frigate-logs.txt

[NickM-27]

the error refers to /tmp/cache, we have seen one user in the past who ran into the error despite it not actually being full, can't remember what the solution was

[FeatherKing]

inside the container i am able to resolve this fqdn right now (i havent restarted the container yet). So it couldnt resolve, filled /tmp/cache and now in a weird state where some of the cameras are kind of working but most arent

root@21bff9818ca1:/opt/frigate# curl fcm.googleapis.com -v

• Trying 216.239.36.57:80...
• Connected to fcm.googleapis.com (216.239.36.57) port 80 (#0)
  GET / HTTP/1.1
  Host: fcm.googleapis.com
  User-Agent: curl/7.74.0
  Accept: /