Closed stevenbrown1862 closed 1 year ago
More info will be needed, if the GPU is hanging then there is some issue on the host most likely. If you are using the integration then /api/stats gets called every 60 seconds so that does not really mean anything
Thanks for the quick response Nick. What more info will be helpful ? When you say issue on the host then are we talking about hardware (bad RAM etc.) or with the operating system ?
More info will be needed, if the GPU is hanging then there is some issue on the host most likely. If you are using the integration then /api/stats gets called every 60 seconds so that does not really mean anything
Looks like https://gitlab.freedesktop.org/drm/intel/-/issues/673.
What is your Linux kernel version? uname -r
?
Have you installed the latest firmware-linux
package (this includes non-free i915 driver)
Looks like https://gitlab.freedesktop.org/drm/intel/-/issues/673.
What is your Linux kernel version?
uname -r
?Have you installed the latest
firmware-linux
package (this includes non-free i915 driver)
I had a similar thought about the kernel after posting the issue so have updated to the latest kernel that Linux mint offered me which is 5.19.0-35-generic. Going from 5.15.0-67.
How would I install the "latest firmware-linux
package (this includes non-free i915 driver)" ?
Not a Mint user myself, but I guess sudo apt install linux-firmware linux-firmware-nonfree
I've now gone 6 days without a crash since updating the kernel version so it looks like that was the problem. Due to when it started happening and the fact it was always linked to detection/motion events I presume its something to do with the OpenVINO detection and the GPU driver in that version of the kernel but thats just an uneducated guess. Thanks for the suggestion @javydekoning .
Thanks for coming back and updating. I'll go ahead and close this then, feel free to create a new issue if something else comes up
@stevenbrown1862 To which kernel version did you upgrade? I'm using Ubuntu 22.04 and had kernel 5.15.0-67-generic and was experiencing the same problem. I tried upgrading to v5.19.17-051917 and now die PCIe Coral is not found anymore...
@stevenbrown1862 To which kernel version did you upgrade? I'm using Ubuntu 22.04 and had kernel 5.15.0-67-generic and was experiencing the same problem. I tried upgrading to v5.19.17-051917 and now die PCIe Coral is not found anymore...
I'm using 5.19.0-35-generic and so far so good...7 days without a crash. I had assumed my issue was related to me using the GPU for the OpenVino detector which presumably you aren't using given that you have a Coral.
@stevenbrown1862
I just changed to kernel v5.15.104 and this one does work with my Coral. Maybe a stupid question, but can I just upgrade to v6.2.8? Or how do I decide which version to choose to upgrade to?
Describe the problem you are having
Hello. I don't know if anyone has any thoughts or have seen similar but I'm running in docker on an i3 6100t and the system crashes every day or so...I initially assumed it was a hardware issue but then I started noticing the following popping up repeatedly in the system log.
Mar 19 13:52:13 steve-HP-EliteDesk-800-G2-DM-35W kernel: [164137.340296] i915 0000:00:02.0: [drm] Resetting rcs0 for preemption time out Mar 19 13:52:13 steve-HP-EliteDesk-800-G2-DM-35W kernel: [164137.342911] i915 0000:00:02.0: [drm] GPU HANG: ecode 9:1:8ed97ff2, in frigate.detecto [475124].
I've had a look in the portainer logs to see if I can see anything relevant and not seen anything obvious. Tonights crash had loads of lines that looked like this just before the crash but not sure whether its relevant:
2023-03-21 19:19:01.876039553 192.168.1.191 - - [21/Mar/2023:19:19:01 +0000] "GET /api/stats HTTP/1.1" 200 2844 "-" "HomeAssistant/2023.3.5 aiohttp/3.8.4 Python/3.10" "-"
Most of the crashes seem to happen when an event is being logged (although there's no indication of an event on tonight's crash).
Version
Beta 10
Frigate config file
Relevant log output
FFprobe output from your camera
Frigate stats
Operating system
Debian
Install method
Docker Compose
Coral version
CPU (no coral)
Network connection
Wired
Camera make and model
Various generic chines, Ezviz and 2 nest cameras
Any other information that may be helpful
No response