We need to get our boxes into atlas to make it easier for people to pull down our vagrant boxes. lala/KB have got a few centos boxes in there. Let's get Fedora in there too.
I will work with lala to add boxes and co-maintain. We'll use this ticket as an update on the progress.
I just created an atlas account to see how tough it was to add an image -- it was super easy. Lala reserved the fedora namespace. Why don't we go ahead and put up images from here http://dl.fedoraproject.org/pub/alt/fedora-atomic/images/testing/vagrant/ for vbox and libvirt?
Lala added me to the admin group for the fedora namespace, and I added the latest libvirt and vbox vagrant boxes from http://dl.fedoraproject.org/pub/alt/fedora-atomic/images/testing/vagrant.
They're at https://atlas.hashicorp.com/fedora/boxes/atomic-host w/ version 22.98, which is the version number reported by atomic host status.
atomic host status
Does that work for us? If so, I'll fill in the description fields and such...
I'm really keen to get this in place for Project Atomic "getting started" purposes:
vagrant init fedora/atomic-host && vagrant up && vagrant ssh
"vagrant init fedora/atomic-atomic" looks fine to me. This means we will have just one atomic vagrant box supported through Fedora cloud SIG. Which is correct if we follow 2 weeks release cycle.
Lala did you mean "fedora/atomic-host"? "fedora/atomic-atomic" seems awkward to me. I'd prefer "fedora/atomic-host" unless there is a reason "atomic-atomic" is better.
Jason, thanks for looking at this. Should we be careful about releasing "testing" content though? Would it be better for now to just put up the actual released version of atomic for F22?
I'll be putting up the released fedora vagrant box for F22 cloud base next week.
Ah my bad. Yes, I mean "vagrant init fedora/atomic-host"
You will need to get with legal and make sure that they have no issues with whatever legal agreements they have in order to do so.
Replying to [comment:4 dustymabe]:
Good point -- we should put whatever the fedora-atomic is here. The lame thing about the official vagrant image is that it isn't updated. Maybe it'd be better to put f22 there now, and, like, does the two-week eventually become the main release?
OK, I deleted version 22.98 (the 2 week testing one) and replaced it with version 22.75 (what you get from the getfedora.org site). If/when the two week becomes the official image, we can update it.
One more thing... here's an interesting/confusing tidbit:
Key components from the official fedora atomic, once updated:
kernel-4.1.6-200.fc22.x86_64 docker-1.7.1-8.gitb6416b7.fc22.x86_64 etcd-2.0.13-2.fc22.x86_64 kubernetes-1.1.0-0.5.gite44c8e6.fc22.x86_64 cockpit-*-0.67-1.fc22.noarch cloud-init-0.7.6-3.fc22.x86_64 ostree-2015.6-2.fc22.x86_64 atomic-1.1-1.git5f631c8.fc22.x86_64
Key components from the testing 2-week release:
kernel: 4.1.6-200 docker: 1.7.1-8 etcd: 2.0.13-2 kubernetes: 1.1.0-0.5 cockpit: 0.67 cloud-init: 0.7.6 ostree: 2015.6-2 atomic: 1.1-1
They appear to be all the same -- so the testing release effectively is the official release?
so I think the strategy here (as discussed in the cloud meeting (09-16-2015)) is to fix a few problems with our vagrant boxes (see https://fedorahosted.org/cloud/ticket/109) and get the F23 vagrant box right and let that be the first vagrant box (+atomic) that we ship in atlas.
The work is in place to index our boxes in vagrant. I have indexed them under my account for now:
https://lists.fedoraproject.org/pipermail/cloud/2015-October/006045.html
On release day I will move them over to the fedora namespace and make a blog post about them.
They are there now:
https://atlas.hashicorp.com/fedora/
We still need to provide appropriate integration with releng. This is covered by the following ticket:
https://fedorahosted.org/cloud/ticket/133
Login to comment on this ticket.