On 01/31/2013 03:35 AM, Ross Brattain wrote:
On Wed, 30 Jan 2013 14:28:28 -0200
Lucas Meneghel Rodrigues <[email protected]> wrote:
This is an old 'feature' [1] that we've implemented in the
hopes of easing SSH key setup on test boxes.
I am using this to bootstrap our systems after anaconda tests.
Oh, I see. Internally we've been using kickstart snippets to add the
needed keys to the clients, so it's been a *while* that I don't use
this. Would it be possible for you guys to adopt a similar solution? I'm
not sure how exactly you are using Anaconda [kickstart or something that
automates interaction with anaconda], so it's always good to ask.
Well, turns out that most of the trouble to set up
an SSH key on a test box is gone with ssh-copy-id.
The 'feature' was also cumbersome to use:
1) Run autoserv once with --ssh-pass, to perform it
2) Run autoserv without it, otherwise it'd break
I usually run with --ssh-pass without issue. However I haven't looked at the
code recently.
Hmm, OK, turns out I had fixed that limitation, nevermind... the code
for the feature haven't changed since, say, 2007 or so.
Also, we ended up bundling pexpect and pxssh for that
reason, which makes our downstream packages to cringe.
ops, a typo, packages/packagers
Having an expect library is useful in general. We are Exscript integrated into
autotest to control our Cisco and Brocade switches during FCoE tests.
Sure, but the problem here is to bundle other upstreams (pexpect and
pxssh by this time should be dead I suppose). Fedora hates bundling, so
my idea was:
- Ok, we have these two bundles (pexpect and pxssh)
- Only one functionality uses it (ssh key setup)
- The feature is mostly superseded these days
(ssh-copy-keys/kickstart), and I guessed nobody was using it.
So that's what happened. I guess I'll have to get back to the drawing
board and withdraw this patch for now, thanks for the feedback.
_______________________________________________
Autotest-kernel mailing list
[email protected]
https://www.redhat.com/mailman/listinfo/autotest-kernel