pciutils in minimal ec2 image

[Bill Nottingham]

This is an explicit addition in the kickstart. Anyone recall *why*?

(If it's important enough, it should just get shoved into core. If it's
not important enough, it should just get dropped....)

Bill

[Matthew Miller]

On Fri, Oct 12, 2012 at 09:10:54AM -0400, Bill Nottingham wrote:
>  This is an explicit addition in the kickstart. Anyone recall *why*?
>  (If it's important enough, it should just get shoved into core. If it's
>  not important enough, it should just get dropped....) 
I don't know the original reason.

I assume it's there because many people have scripts which use the output to
determine if they're running virtual or local. SUSE's ec2 utils use it to
discover if there's an nvidia graphics card, which I don't think we're
concerned with.


-- 
Matthew Miller  ☁☁☁  Fedora Cloud Architect  ☁☁☁  <mattdm(a)fedoraproject.org&gt;

[Bill Nottingham]

Matthew Miller (mattdm(a)fedoraproject.org) said: 
>  On Fri, Oct 12, 2012 at 09:10:54AM -0400, Bill Nottingham wrote:
>  > This is an explicit addition in the kickstart. Anyone recall *why*?
>  > (If it's important enough, it should just get shoved into core. If it's
>  > not important enough, it should just get dropped....)
>  
>  I don't know the original reason.
>  
>  I assume it's there because many people have scripts which use the output to
>  determine if they're running virtual or local. SUSE's ec2 utils use it to
>  discover if there's an nvidia graphics card, which I don't think we're
>  concerned with. 
Discover if there is an nvidia graphics card *on their EC2 image*? There
must be something obvious here that I'm missing.

If it it's for virt/phys determination, I'd be tempted to just put virt-what
in core.

Bill

[Jon Stanley]

On Fri, Oct 12, 2012 at 9:31 AM, Bill Nottingham <notting(a)redhat.com&gt; wrote:

>  Discover if there is an nvidia graphics card *on their EC2 image*?
> There
>  must be something obvious here that I'm missing. 
EC2 supports GPGPU instances, so it's quite possible you'd have an
nvidia card in your instance.

[Matthew Miller]

On Fri, Oct 12, 2012 at 09:31:32AM -0400, Bill Nottingham wrote:
>  > I assume it's there because many people have scripts which
> use the output to
>  > determine if they're running virtual or local. SUSE's ec2 utils use it to
>  > discover if there's an nvidia graphics card, which I don't think we're
>  > concerned with.
>  Discover if there is an nvidia graphics card *on their EC2 image*? There
>  must be something obvious here that I'm missing. 
Ayup. Amazon has Cluster GPU Instances, which have Fermi M2050 GPUs. This is
a super-huge selling point for HPC cloud, because building a local Tesla
cluster is somewhat painful. (I say from experience.)

>  If it it's for virt/phys determination, I'd be tempted to
> just put
>  virt-what in core. 
+1 from me. But I don't know for sure the original reason was; I'm just
guessing.

-- 
Matthew Miller  ☁☁☁  Fedora Cloud Architect  ☁☁☁  <mattdm(a)fedoraproject.org&gt;

[R P Herrold]

On Fri, 12 Oct 2012, Bill Nottingham wrote:

>  This is an explicit addition in the kickstart. Anyone recall *why*?
>
>  (If it's important enough, it should just get shoved into core. If it's
>  not important enough, it should just get dropped....) 
no specific recollection, but almost certainly reflective of 
the fact that RPM will not identify a dependency, and so pull 
in content from the Turing complete environment, of %pre and 
%post scripts

-- Russ herrold