#45 drop Workstation install tree?
Closed: Fixed 3 years ago by catanzaro. Opened 4 years ago by kevin.

Greetings.

In releng https://pagure.io/releng/issue/7403 it has been suggested that we stop making a Workstation install tree. This would mean that users could no longer use the Workstation boot.iso or pxe files to install Workstation. They would need to use the normal boot.iso or pxe files and pick and choose the groups/packages out of Everything in order to install a workstation.
There is talk about adding a option to the installer so you could pass the product to it and get that product by default, but this has not been implemented yet.

So we could:

  1. Just keep making it.
  2. Wait for a way to easily pass to the installer the product so defaults are set, then drop the tree.
  3. Just drop it now (f29 I assume)

Let us know what you would like to do, or if more information is required.


As I understand it, right now, the only practical difference would be that there wouldn't be a Workstation-branded netinstall, and that to install Workstation via that method, one would need to add a command-line boot option?

I'm generally in favor (the different trees are part of exploration for a path not taken) but I do think we should make Workstation Netinstall easy in some fashion.

I just wanted to add one point of clarity here.

This would mean that users could no longer use the Workstation boot.iso or pxe files to install Workstation.

Right now the only difference between the boot.iso for Everything and the boot.iso for Workstation is a single text file in the iso telling anaconda which installclass to use. This installclass can be chosen using a kickstart (non-interactive) today. In #2 as kevin mentioned above we are proposing for the interactive case we just make it so the user can select another option in the "Everything" boot.iso to make it think it's a workstation boot.iso

The things we need to deliver for the workstation are a) a live image and b) a regular install iso.

I don't think we need a netinstall or an everything dvd or any other alternative installation methods.
I at least would welcome if we stopped producing anything other than the necessary 2 deliverables.

But no, making an "everything iso where the user can manually select "workstation" does not fit the bill.

@mclasen cool. What exactly is a regular install iso? Does that have an embedded repo on it?

I was mistaken. I just want the live iso. But I think this will need to be discussed in the wg - I already got some dissenting opinions on this.

I actually think having one unambiguous, marketed deliverable is important. I mean if we have ambassadors passing media out at conferences, the ambassadors should be passing out workstation live images, instead of say a pick your own adventure dvd installer, or whatever. maybe the netinstall image is useful for certain usecases, but it's not something we should be marketing. of course, we also have to think about the messaging for atomic, if we tech preview that. We need to make it clear what stuff is the product deliverable, and what stuff is on the side, imo. So, just to be clear, if netinstall is properly de-emphasized, i don't think it matters if if's workstation specific or not. since it's not what we're expecting users to use.

If you guys don't want a branded netinstall, I'm cool with that. We should document how to do a Workstation netinstall from the generic image -- this is useful for computer lab / classroom use cases, and I assume office environments as well. Plus if someone wants to install in a VM, it's handy. But sounds like we don't need a separate deliverable.

well... if we only produce one workstation iso variant, that is the way we want people to install the workstation. its no good to say 'we only produce and qa the live image' and then tell every classroom to install in some other way.

But a live image is not a workable way to install even a dozen machines. I think these cases apply to both the "student" and "enterprise developer" personas in the PRD (and I assumed goes hand-in-hand with work like Fleet Commander.)

If the WG is clear that this isn't a use case Workstation wants to support, I'll support that, though.

But a live image is not a workable way to install even a dozen machines. I think these cases apply to both the "student" and "enterprise developer" personas in the PRD (and I assumed goes hand-in-hand with work like Fleet Commander.)

In those cases, I think we can expect the person performing the installation to be familiar with whatever method we have documented for performing a netinstall, right? It should be easy to do, like "select the Workstation group in Anaconda."

Metadata Update from @catanzaro:
- Issue tagged with: meeting

3 years ago

Discussed in WG today -- I need to check with someone in the Anaconda team to find out how big a change the installclass bits are, since we'd like those before getting rid of this tree.

Metadata Update from @pfrields:
- Issue assigned to pfrields

3 years ago

Discussed in WG today -- I need to check with someone in the Anaconda team to find out how big a change the installclass bits are, since we'd like those before getting rid of this tree.

Does that mean you are going to open the RFE mentioned here ?

Metadata Update from @ryanlerch:
- Issue set to the milestone: Fedora 30

3 years ago

Metadata Update from @ryanlerch:
- Issue untagged with: meeting

3 years ago

This change was made in anaconda:
https://github.com/rhinstaller/anaconda/pull/1716

Does this clear the way for us to do away with the WS tree? What needs to happen next?

Metadata Update from @pfrields:
- Issue tagged with: meeting

3 years ago

Metadata Update from @pfrields:
- Assignee reset

3 years ago

Metadata Update from @pfrields:
- Issue assigned to pfrields

3 years ago

Yes, just needs config changes now I think...

Metadata Update from @ryanlerch:
- Issue untagged with: meeting

3 years ago

Closing this. Kevin, our response is that we don't need the install tree.

Metadata Update from @catanzaro:
- Issue close_status updated to: Fixed
- Issue status updated to: Closed (was: Open)

3 years ago

Login to comment on this ticket.

Metadata