>> Hi all,
>>
>> As some of you may know, we've been working on adding a feature to
>> Foreman to allow you 'discover' hardware, and then consume it at later
>> date. This follows a process where:
>
> So what 'state' is such a discovered server presumed to be in
> as far as Foreman is concerned?
Any discovered hardware is presumed unused and availble for
deployment. Obviously, any machine already known to Foreman will have
a TFTP boot file for Localboot, and therefore will not end up in the
discover image.
However, by default the discovery ramdisk makes no changes, so if you
have hardware which is in-use, but not yet know to foreman, it's non
destructive - simply add the host to foreman, disable build, and
reboot the server, it should come back up just fine.
Obviously right now this is aimed at a network fully controlled by
Foreman. I've got some thoughts about how to handle heterogeneous
networks for the next version 
>> Hardware is booted into a small ramdisk environment
>> Ramdisk registers with Foreman
>> Ramdisk waits for further instructions
>> Admin chooses a piece of hardware to be provisioned with an OS
>> Hardware reboots and installs the chosen OS
>
> This sounds pretty interesting, but could you please elaborate
> how it is different from Foreman owning that said server the minute
> it appears in the datacenter? You don't have to do anything with
> it as far configuration is concerned, of course, but other than that
> it'll be as first class citizen as any other piece of hardware.
(MFridh might weigh in here, this was his idea orginally…) You're
right, ofc, but if you've just received a lot of hardware, entering
all of it into Foreman is time-consuming and error-prone - I hate
typing MACs in manually, and that's made worse if you're relying on
remote hands to do the racking and send you the MACs.
Consider the situation where 50 servers have arrived at your
datacenter. You've racked them up, but you're not sure yet what you're
going to use all of them for. So, you switch them all on (with no
pre-registration of data in Foreman) and then they boot to Discovery,
and register themselves in Foreman. Now, at some later time, you can
think "Oh, I need a new webserver", go to the Discovered Hosts, browse
the Fact data to find one with, say, enough memory or disk space for
the role, and then provision it. Also, when you delete a machine from
Foreman, you can reboot it, and it goes back into the pool of
Discovered Hosts for re-use.
It's also useful as a pre-install environment to apply BIOS/Firmware
updates, or do Hardware Raid configuration, or anything else which
would be difficult to squeeze into the installation environment.
In all, what you suggest is correct, I just feel the flow if this is
slightly more natural, especially if machines are changing roles or
hardware is being shared among teams.
Hope that helps,
Greg
···
On 14 January 2013 04:53, Roman Shaposhnik wrote:
> On Thu, Jan 10, 2013 at 7:39 AM, Greg Sutcliffe > wrote:
