[HW Accel Support]: VA-API iHD driver fails to initialize on kernel 6.17 — Intel Arc A310 (glibc mismatch) #22748
-
Describe the problem you are havingAfter upgrading my host kernel from 6.8.0-106-generic to 6.17.0-19-generic, the bundled The same Frigate version and config worked perfectly on kernel 6.8. The host's package repos (Intel PPA This is likely to affect an increasing number of users since kernel 6.17 is needed by other software (e.g., Immich OpenVINO requires 6.17 for Arc GPU detection). Version0.17.1-416a9b7 Frigate config filemqtt:
enabled: true
host: <REDACTED>
user: <REDACTED>
password: <REDACTED>
ffmpeg:
hwaccel_args: preset-vaapi
go2rtc:
streams:
# 6 Dahua cameras via RTSP on a separate VLAN
cameras:
walkway:
enabled: true
ffmpeg:
inputs:
- path: rtsp://<REDACTED>
roles:
- record
- detect
# ... 5 additional cameras with similar config
detect:
enabled: true
detectors:
ov:
type: openvino
device: GPU
model:
width: 300
height: 300
input_tensor: nhwc
input_pixel_format: bgr
path: /openvino-model/ssdlite_mobilenet_v2.xml
labelmap_path: /openvino-model/coco_91cl_bkgr.txt
record:
enabled: true
semantic_search:
enabled: true
face_recognition:
enabled: true
lpr:
enabled: true
version: 0.17-0docker-compose file or Docker CLI commandversion: "3.9"
services:
frigate:
container_name: frigate
privileged: true
restart: unless-stopped
stop_grace_period: 30s
image: ghcr.io/blakeblackshear/frigate:stable
shm_size: "1940mb"
devices:
- /dev/dri/renderD128:/dev/dri/renderD128
- /dev/dri/card0:/dev/dri/card0
volumes:
- /etc/localtime:/etc/localtime:ro
- ${DOCKER_FOLDER}/config:/config
- recordings:/media/frigate
- type: tmpfs
target: /tmp/cache
tmpfs:
size: 1000000000
ports:
- "8971:8971"
- "5001:5000"
- "8554:8554"
- "8555:8555/tcp"
- "8555:8555/udp"Relevant Frigate log output-na-Relevant go2rtc log output-na-FFprobe output from your camera-na-Install methodDocker Compose Object DetectorOpenVino Network connectionWired Camera make and model-na- Screenshots of the Frigate UI's System metrics pages-na- Any other information that may be helpfulNo response |
Beta Was this translation helpful? Give feedback.
Replies: 4 comments 1 reply
-
|
This is a kernel/driver compatibility issue outside of Frigate's direct control. The Frigate container is built on Debian Bookworm with glibc 2.36, and the newer Intel media driver (26.1.4) requires glibc 2.38, which creates an incompatibility when running on kernel 6.17. (1) Your configuration for Intel hardware acceleration with VAAPI and OpenVINO on GPU is correct based on the documentation: ffmpeg:
hwaccel_args: preset-vaapi
detectors:
ov:
type: openvino
device: GPUPotential workarounds to consider:
detectors:
ov:
type: openvino
device: CPU
The documentation notes that for Intel GPUs, the default driver is This appears to be a container base image limitation that would need to be addressed upstream by the Frigate project. 📚 Sources: Have another question? Just tag @inkeep. |
Beta Was this translation helpful? Give feedback.
-
|
Is this specific to latest Immich version? That's not a previous requirement as that kernel is relatively new. Also worth noting that Intel Arc integrated Graphics (Core 120H) is working correctly for me on linux 6.19 in the current Frigate container. We did update the driver in 0.18 so it is possible this can be improved in your case from that, otherwise there may be something else that is incorrect. |
Beta Was this translation helpful? Give feedback.
-
|
I may have to blow away the entire stack and try again but for Immich smart search it was erroring out on Intel drivers until I upgraded to the latest firmware. What kernel did you test on so I can match?
…-------- Original Message --------
On Friday, 04/03/26 at 16:14 Nicolas Mowen ***@***.***> wrote:
Is this specific to latest Immich version? That's not a previous requirement as that kernel is relatively new. Also worth noting that Intel Arc integrated Graphics (Core 120H) is working correctly for me on linux 6.19 in the current Frigate container.
We did update the driver in 0.18 so it is possible this can be improved in your case from that, otherwise there may be something else that is incorrect.
—
Reply to this email directly, [view it on GitHub](#22748?email_source=notifications&email_token=ABVHSJBUSEBVS6PJW7CX6LT4UALSDA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCNRUGQZDEMBZUZZGKYLTN5XKMYLVORUG64VFMV3GK3TUVRTG633UMVZF6Y3MNFRWW#discussioncomment-16442209), or [unsubscribe](https://github.com/notifications/unsubscribe-auth/ABVHSJEUFE624TL5VJNOMB34UALSDAVCNFSM6AAAAACXMDIYR6VHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTMNBUGIZDAOI).
You are receiving this because you authored the thread.Message ID: ***@***.***>
|
Beta Was this translation helpful? Give feedback.
-
|
Bizarre but it was a Proxmox 9.x issue I found the fix here . I ended up blowing away and rebuilding from a stock Debian Trixie cloudinit image with stock drivers. |
Beta Was this translation helpful? Give feedback.
Bizarre but it was a Proxmox 9.x issue I found the fix here . I ended up blowing away and rebuilding from a stock Debian Trixie cloudinit image with stock drivers.