[chef] Re: Re: Ohai convention for hardware details?


Chronological Thread 
  • From: "steve ." < >
  • To: " " < >
  • Subject: [chef] Re: Re: Ohai convention for hardware details?
  • Date: Mon, 30 Jun 2014 16:58:21 -0700

That's great - it looks like Windows and Linux use the same behavior, but Darwin apparently can't use dmidecode as it doesn't expose the necessary memory regions.

Not to mention other operating systems...


On Mon, Jun 30, 2014 at 2:57 PM, Maxime Brugidou < " target="_blank"> > wrote:

We run dmidecode on windows using dmidecode.exe at http://gnuwin32.sourceforge.net/packages/dmidecode.htm

This actually works very well. We have a custom ohai plugin that solves https://tickets.opscode.com/browse/OHAI-333 (I don't know why we didn't do a PR yet, I'll check).

DMI information is the standard for hardware discovery IMHO we should keep using it.

On Jun 30, 2014 11:43 PM, "steve ." < " target="_blank"> > wrote:
So there's a pretty standard place for people to look for information about a node's hypervisor characteristics (is it VMware?  Openstack?  Cloudstack?).  I'm thinking of cloud_provider.

And we agree that it's so important to know what kind of platform we're on that we have platform_family.

But there's not really a standard for figuring out what our hardware is.

On Linux, we run dmidecode, cover our eyes and jam that into one structure.

On OS X, we run system profiler, cover our eyes and jam that into a different structure.

On Windows, _________________.  (I haven't tried yet.  I'm afraid to.)

I think this is going to get more important as people use Chef on different kinds of physical hardware.  It would be great if there was some kind of standard for hardware that Ohai plugins on different platforms attempted to conform to - something like:

'hardware': {
  'manufacturer': 'Falken Robotics',
  'model': 'WOPR',
  'revision': '01'
}

Is anyone interested in/already working on this?  Is it on someone's roadmap already?  Or did someone try working on this and give up because ... ?  




Archive powered by MHonArc 2.6.16.

§