double-take: [BUG] Won't connect to DeepStack

SOLVED Solution provided by @szmidtpiotr works! 😃 https://github.com/jakowenko/double-take/issues/167#issuecomment-975376875

Describe the bug I have installed DeepStack as a seperate Docker container. Forwarded port 5000 to 4999 (on the host). This used to be port 80, but I thought some conflict could’ve been the issue, which it wasn’t. I am using a key. Double Take won’t connect to my DeepStack. Frigate works fine.

Version of Double Take v1.6.0-893328f

Expected behavior Double Take should connect to my DeepStack detector If there are any errors, they should be shown in the logs.

Screenshots image image image

Hardware

  • OS: Home Assistant Operating System
  • Browser: Firefox

Additional context

# Double Take
mqtt:
  host: 192.168.10.105
  user: <USER>
  password: <PASS>

frigate:
  url: http://192.168.10.105:5000

detectors:
  deepstack:
    url: http://192.168.10.105:4999
    key: <KEY>
docker run -d --restart=unless-stopped --name="deepstack" -e VISION-DETECTION=True -e VISION-FACE=True -e API-KEY="<MY KEY>" -v localstorage:/datastore -p 80:5000 deepquestai/deepstack

Tested both with and without key.

About this issue

  • Original URL
  • State: open
  • Created 3 years ago
  • Reactions: 4
  • Comments: 56 (2 by maintainers)

Most upvoted comments

Just in case anyone else stumbles on this thread and gets as far down as this, I am running DS on a Proxmox VM, with Intel GPU passthrough, however the fix, as other mention is adding VISION-FACE=True as an environmental variable but for me the key part is ‘True’ is case sensitive! I had it first as ‘TRUE’ and then as ‘true’ and DT kept showing the red not connected. I then changed it to ‘True’ and DT now shows connected to Deepstack. Now time for a play

The solution definitly worked 😃

image I did need to restart my Proxmox system after adding the config as I was unable to reload the kvm_intel module.

  1. Check if nested virtualization is enabled on the Proxmox server
cat /sys/module/kvm_intel/parameters/nested
N
  1. Activate the parameter
echo "options kvm-intel nested=Y" > /etc/modprobe.d/kvm-intel.conf
  1. Activate virtualization on guest
qm set 104 --cpu host # Update VM ID to match yours

Or via Web Ui: image

  1. Reboot Host
cat /sys/module/kvm_intel/parameters/nested
# Y

Enabling hv_evmcs caused an error in my setup. As I don’t fully understand the effect of this option, I won’t fiddle with it too much 😉

image

https://www.reddit.com/r/homeassistant/comments/qi53h5/comment/hjz4ci7/?utm_source=share&utm_medium=web2x&context=3

Bug is related to Virtualiization of the CPU as i understand. If you use proxmox to setup HASSIO then simply change CPU type to HOST in your HA vm. Helped for me

I have been trying to get Deepstack on a Jetson Tx2 4GB which is close to the Jetson nano as far as performance. @LordNex it was crashing when connecting. I found it was because of memory pressure so increased swap size on connected SSD and it stopped it crashing.

@Heisenberg2980 But that is for a VM specifically, I can’t recall yet how to do it for a container. I don’t think that’s possible, because containers always see the host CPU as it is.

cpu

To be honest i dont know. Im not an expert, just found that sollution in other place, that worked for me - but if you could test and give us update that would be grate.

My fix for this issue was adding the environment variable VISION-FACE=True in my docker compose file for Deepstack.

Same issue here, running a LXC Proxmox container, Deepstack won’t connect, but using a VM, with the ‘host’ type CPU configured works fine.

I would like to use an LXC, because of lower resources…

Same issue here, Double Take cannot connect to Deepstack running on LXC Proxmox container returning the error “Face endpoints not activated”.

@Calimerorulez I just ran command lscpu | grep avx in both the container and the host, and it returns blank, so I guess my server doesn´t have AVX capabilities, and if that is a requirement for Double Take to find Deepstack, that would explain why it is not connecting

Exact same issue here. Running in seperate dockers on the latest unraid.

I even tried moving my frigate port to 5001 leaving deepstack at 5000. Still no luck.