+1 as long as rel-eng is OK with it. If it's just a "need a person to do it" issue, I'd suggest someone from the SIG helps rel-eng.

How are the updated images tested? Do they get the same level of scrutiny as the "release" version?

@jdoss can you address the question about testing?

@jdoss what are the cloud images that are under consideration here? I would prefer a specific list to be able to do an informed decision. Will the GA images still be available or will there be only updated images?

Sorry for the delay in response folks. For some reason Pagure doesn't subscribe you to an issue when your name is linked in the description.

@zbyszek In my initial talk with @dustymabe about getting more frequently updated cloud images to all of the supported providers, he did have @kellin and @mohanboddu looped in from rel-eng about having some testing created to ensure we have a basic functionality of the images that are being updated post GA.

@till All supported provider images are under consideration. Leaving the GA image on each provider is idea if they support many images under our org. For example, we don't want to yank the GA AMI off of our AWS account as people could be using it. Providing updated information on https://alt.fedoraproject.org/cloud/ with the updated AMIs IDs for reach region and updated base images was the initial plan. From a user experience an security standpoint, we should push them towards the updated images.

Since the vote is currently at (+1,0), I suggest this change be discussed at today's meeting.

We decided to hold the next meeting on the 20th because of flock travel. Let's discuss this then.

+1 to this from me. It will need a bit of releng work and a bit of QA work, but I think thouse can be found.

I'm generally very positive about the change, but I worry about the potential havoc if images that were not bootable or otherwise faulty were pushed out. I see that testing is planned. Can we get some specifics on what those tests would be and how much work they would required to set up?

There has been cases where the GA image isn't usable by end users. Fedora 26 shopped a busted rpcbind package which made using the Cloud Vagrant image unusable if you use NFS. Fedora 28 shipped a kernel when booted on Digital Ocean, would take 5 to 10 minutes to boot. These issues were not solved until the next GA or without manual intervention to release a new one.

Even if we happen to ship an image that is causing end user pain, having the mechanisms in place to turn around and ship an updated base image will be better than what we have now.

As of testing, the idea is to start with the basics. Will it boot. Can we install some packages. If those are not enough we can add in more tests to cover more cases. If we start out trying to get 100% test coverage we won't hit our goal of getting updated images out the door.

+1 to the change. But few considerations:

Someone needs to test the images, either FedoraQA or the SIG(most likely).
RelEng can release the images but RelEng has to work with infra in figuring out where to release them. (Just to make a decision)
Someone from SIG should help RelEng to automate the release process. (RelEng is currently dont have enough resources)

Approved in today's FESCo meeting

This was discussed in yesterday's FESCo meeting (2018-08-20):
AGREED: F29 Change: Cloud Provider Image Updates - Approved by in-ticket votes (+1: 6, -1: 0, +0: 0)