I have been tinkering with resin and have at least gotten as far as getting my test NUC onboard - cool!
What I am trying to figure out is if the following is possible. I am open to hacks and/or meatball programming.
I will have a fleet of NUCs out in the world.
They will be running a very heavy set of applications and are based on RHEL7.1. Would it be possible to either splice in the resin management services to my red hat installation or is there a way - perhaps using vagrant or something to delpoy my enviroment? I have been googling like crazy and see an article on how to create a custom base - but it seems to be directed to ARM machines…
I’m not sure what the requirements of you application are, but because the NUC is x86 you should be able to swap out your resin base image with most docker hub images. I found this: https://hub.docker.com/r/gang929/rhel7.1/
Let me know if it works for you,
I think what i really mean is can I replace what is called the Device OS with something else like RHEL 7.1. I have tried to dockerize our application stack and its just too complicated for my pea brain.
I am trying to study Yocto to see what I can do there - If I could run vagrant directly on the Device OS with my stack in that - it would work. Or if I used RHEL and added the resin stuff for the management and provisioning it would be great as well…
Could you give a bit more info about why you want to replace the device OS? In general, resin containers are very powerful. If you make a container based on the RHEL base image, you should have the same experience and power you have now, with all the benefits of resin. Is there a particular objective you feel won’t be met using this approach?
This is really about my ignorance of docker i think and an impatience to get things going. Our software stack is quite large and currently relies on systemd to get everything started. It has a number of C++ daemons as well shell scripts, tomacat and mariadb.
I could disect it - but right now I have an ISO that installs everything correctly and starts up. Its pretty fragile in its current state to make things worse
An ideal solution would be to be able to run a vagrant machine in the container - I havent see if that is possible. It would be a big stick approach that would allow me to move ahead …
Our standard container images all have systemd inside the container, (you can turn it on by writing simply “ENV INITSYSTEM on” in the dockerfile) and you’re off to the races. It should really be the fastest way to get going. they are however debian images, so if you want a RHEL image, you may want to look at things like – http://jperrin.github.io/centos/2014/09/25/centos-docker-and-systemd/ for a centos image. Other resources exist for RHEL, just a google away
Thanks for the help! I am starting to make good progress. One (more) question - when you run bitbaker what is the target reciepe i should be using?