You have rel-eng and QA buy in for this?

I'm especially interested in the QA side of things given the failures in F-23 that I've seen no follow up from in the retrospective ticket

I've talked with dgilmore about it. The images are being built already. As far as QA buy-in the only contact from there that I really interface with is roshi, who has been a part of the discussions in the past. Should we formally reach out to adamw or someone else to have a conversation around this?

I have several concerns here that I suspect will be echoed by the rest of QA:

• Who's signing up to make sure that enough testing gets done? This can't be just roshi

• Assuming that the testing follows a process similar to the releases, what format are the test matrices goign to take on the wiki? How should they be named?

Bringing the topic up at the next QA meeting would be a good way to approach getting QA buy-in.

I'll be at the meeting on Monday at 16:00 UTC. If we can't get QA buy in then we won't do this.

Dusty

The QA meeting was canceled today.

Last time when we discussed about this we came to a decision about testing the updated images by Cloud WG itself as QA is already overloaded. I will have to find that IRC log (must be somewhere).

In the Fedora 22 cycle I had created one ticket [1] for the update, but may be it was lost.

[1] https://fedorahosted.org/rel-eng/ticket/6219

As decided in the meeting on 2015-01-06 we will release an update cloud base image this month (hopefully by next week). I will discuss with rel-eng and update the list with the details.

Some information has come to light about the release process around the 2 week atomic image releases. Adam has decided to scrap the existing release process and write it from scratch.

Considering the existing release process is "duct tape and band-aids" we are going to scrap the idea of releasing updated cloud images until the F24 release.

On a side note: Does the recent uncovering of the glibc cve affect our decision that we made not to do a one-off release of Fedora cloud (half way between F23 and F24)?

"''On a side note: Does the recent uncovering of the glibc cve affect our decision that we made not to do a one-off release of Fedora cloud (half way between F23 and F24)?''"

I'd really prefer we stick to the one-off release, and perhaps accelerate it.

I will note that the process that makes the nightly atomic composes also makes all the cloud images. They are there, we just need to release them.

Replying to [comment:10 ausil]:

I will note that the process that makes the nightly atomic composes also makes all the cloud images. They are there, we just need to release them.

Yep. You are right. I brought this up but from what I understand the actual process to get the images into the right download locations and updated on the website was more of a pain than it is worth. I believe Adam is the person who has the most information about this (at least when we were having the discussion).

The discussion we had took place before the glibc vulnerability so it may now be worth the extra work.

We are not doing this till F24 release. Mean while we have to find out the exact steps required to get the updated image out as a release.

A few users have been requesting this as of late: [mail thread] (https://lists.fedoraproject.org/archives/list/cloud@lists.fedoraproject.org/message/IADTY4BL3M3YBL7QKOTYNRR54DOP6XHB/)

@mattdm had some questions in this mail copying them here:

So, we are already making these; you can see them at
https://apps.fedoraproject.org/autocloud/compose — they're currently
under a compose ID with "Atomic" in the name, but AIUI that's going to
be split out. (I was worried that these might not include updates, but
that's fixed.)

Let's make a list of what needs to happen... here's what I can think
of:

1. Decide if we are okay with the level of automated testing these are
   getting, or if we need human testing, or if we need more automated
   testing

2. If we need more automated tests, someone needs to sign up for that

3. If we need human testing, someone needs to sign up to write the
   release criteria

4. And, someone would need to commit to doing the validation every
   time.

5. Work with release engineering and infrastructure to adapt the Atomic
   Host gating/release process for Cloud Base

6. Work with release engineering to get updated images to mirrors and
   stuff

7. Someone needs to sign up to update the Vagrant Atlas index

8. Work with websites to update cloud.fedoraproject.org to also offer
   updated images. (I'm thinking the page should default to the latest,
   but there should be some way to "scroll back" all the way to the GA
   releases.)

9. Decide if we want to switch Cloud Base entirely to this
   automatic process and eschew release milestones

I'm sure I'm missing something -- what else should be added?

Howto proceed here?

I'm fine with the automated testing. Manual testing maybe needed when kickstarts or release changes.

migrated to https://pagure.io/cloud-sig/issue/138