Closed flaviostutz closed 6 years ago
thanks @flaviostutz looks perfect!!
@flaviostutz btw do you run kerberos.io on swarm?
Yes. I run it using Docker stack with docker-compose.yml
Cool, just curious. How many cameras are you processing?
Just one. I have another project that would use Kerberos for 80 cameras, but it is paused now. Before going on with it I need an way to split the camera feed into Kerberos for image processing and the regular tools the existing infrastructure has to minimize the impact of the image recognition I want to leverage to current operations... I coudn't manage to find a splitter that don't do the stream recoding . Each stream costs 8% CPU for recoding (x 80 = 8 full CPUs!) and I wish I could avoid that.
Hmm interesting if you want to change ideas, don't hesitate to contact me. Btw we now have a single image, so you don't need to compose the containers anymore (it's also ARMv7 and ARMv8 supported). http://hub.docker.com/r/kerberos/kerberos/
Great! I didn’t know that. You could deprecate those separate images, don’t you?
So... do you know any rtsp repeater that would avoid reencoding the stream?
Sent from my iPhone
On 5 Jun 2018, at 05:56, Cédric Verstraeten notifications@github.com wrote:
Hmm interesting if you want to change ideas, don't hesitate to contact me. Btw we now have a single image, so you don't need to compose the containers anymore (it's also ARMv7 and ARMv8 supported). http://hub.docker.com/r/kerberos/kerberos/
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.
For using Kerberos on Docker Swarm stacks we need the usage of newer docker-compose, so we need to migrate the semantics of volumes_from to another strategy.