You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The idea is to try out ansible-container for use in TOAD to be able to re-use the roles we have here, and maybe smooth out the portions we have.
Also, I put together a recipe for using ansible-container, and wrote a blog article about if you want to get your feet wet with ansible container. Here's the blog article if you're interested
I'll create a poc/ansible-container branch if anyone else wants to check it out.
I put a 17.05 milestone on it, unsure if this will be sticky or we'll chuck it out later, but, I think there's a possibility it could be useful.
The text was updated successfully, but these errors were encountered:
Hitting some snags running traefik with ansible-container, I can't find a good path for saying "this image is totally a third party image and I don't want to modify it" so... that's causing me a hiccough. Have an issue posted in ansible container @ ansible/ansible-container#400
So I think it might be good to set some boundaries about how far we might want to take this initial step.
Originally, @zshi-redhat requested that we provide some images upstream (Docker Hub for example) that has the base set of files and packages on the image. This would help to reduce instantiation time, and reduce future spin ups since the base image would contain all the data, removing the need to hit upstream package downloading from various sites.
Looks like there might be some issues using things like Traefik and orchestration with ansible-container. Personally, I think as an initial first step moving from a docker-compose based deployment method to an ansible-container based deployment method. To me, that seems like too much change in a short period of time. I'd rather we be trying to stabilize for partners, get our documentation up to date, and refine additional jobs in JJB.
My opinion then is, that we look to see if ansible-container will provide us a simple set of tools, allowing us to re-use our existing Ansible roles and playbooks, to build initial container images for Jenkins, ELK, etc, and push those into the NFVPE namespace on Docker Hub, and update our docker-compose.yml to consume those images.
However, it would also be good to provide a way that you could build the container from scratch, if possible (basically keep the same method as we have now for an override).
I also believe that you will not be able to spin up TOAD directly from the images being deployed, and that you'll still need to run the ansible-playbook command to customize the image. However, it should limit the amount of outside network utilization, and should result in a mostly "offline" style deployment.
At this point I'd like to ask for feedback so that we have a clear direction on how we might want to use ansible-container. I think there is some good work here and that we can leverage it for container building, and publishing that to Docker Hub.
+1 using it for the build pipeline is probably where we can get the most value out of it -- and leverage the work that's already done without having to say -- transfer everything to Dockerfiles.
There may be some duplication of code a little bit between the docker-compose.yml and the containers.yml files, but, it shouldn't be too painful to maintain.
The idea is to try out ansible-container for use in TOAD to be able to re-use the roles we have here, and maybe smooth out the portions we have.
Also, I put together a recipe for using ansible-container, and wrote a blog article about if you want to get your feet wet with ansible container. Here's the blog article if you're interested
I'll create a
poc/ansible-container
branch if anyone else wants to check it out.I put a 17.05 milestone on it, unsure if this will be sticky or we'll chuck it out later, but, I think there's a possibility it could be useful.
The text was updated successfully, but these errors were encountered: