[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: [master f17-branch] The Great Loader Removal

On Tue, 2012-02-14 at 10:30 -0500, Chris Lumens wrote:
> > [If you're wondering what will replace all that code, there will be some
> > patches coming Real Soon Now that add a new 'anaconda' dracut module which
> > does *most* of the same things.
> I can't possibly approve of this patch set hard enough, though I'm
> hesitant to ACK it until I see what's coming next.  Loader had a whole
> lot of little edge case pieces that were important to someone.  I hope
> you've got them all figured out for somewhere.

Short answer is: yeah, working on that, but we don't need 'em for Alpha.

AFAICT there's basically four classes of things loader did:

1) system setup,
2) workarounds for bugs that no longer exist,
3) actual stage 1 tasks (kickstart, driverdisk, repo=, &c),
4) setup for anaconda's environment in stage2

Most of #1 should already be handled by dracut/systemd, and we can
obviously just drop #2.

The current F17 test builds do not have loader in stage1 at all, *but* I
talked to QA about it.. and it seems none of those tasks seem are in the
Alpha release criteria. 

Everything else that's missing we're finding out pretty quick through

But yeah, there's a lot of weird knotty edge cases.
(e.g: if plain 'ks' is specified, you construct the kickstart path from
- DHCP 'next-server' option (or server IP if non-existent), plus
- DHCP 'filename' option (or '/kickstart/' if non-existent), plus
- '$IP_ADDRESS-kickstart' if the constructed path ends in '/'.

And yes, the anaconda dracut module handles that. In less than 20 lines
of bash, even. Patches for that should start appearing Real Soon Now.


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]