[FACT-1827] AIX hardwaremodel (os -> hardware) fact is non-standard Created: 2018/03/02 Updated: 2018/05/22
|Affects Version/s:||FACT 3.9.4|
|Remaining Estimate:||Not Specified|
|Time Spent:||Not Specified|
|Original Estimate:||Not Specified|
AIX 7.1 TL4 (powerpc) - PE 2017.3.4 (facter 3.9.4)
The fact should be fixed to be more inline / compatible with other operating systems.
|Method Found:||Customer Feedback|
|QA Risk Assessment:||Needs Assessment|
Was setting up rspec testing on a module after porting it to AIX and the facts are not being found. Part of the problem is that the default "hardwaremodel" filter is `x86_64`. It does have other OS cases for Solaris (i86pc) and BSD (amd64) for example. While we could add a case for AIX https://github.com/mcanevet/rspec-puppet-facts/issues/69 . The problem is, as I started looking, it looks like the "hardwaremodel" is populated with the actual product name model of the machine (such as "IBM,8284-22A"), while other OS's seem to use something like the processor architecture (x86_64). I think that hardware information belongs in a different fact, like maybe productname (based on what I am seeing on OSX) and the hardwaremodel fact should be "powerpc"
|Comment by Branan Riley [ 2018/03/02 ]|
Thanks for reporting this. We agree that the use of various hardware-related facts (hardware, ISA, product, and certainly others) have not always been consistent. Feedback like this is useful for us to be able to sort through what users expect from the names of these facts, so we can fix it in the future.
Generally, changing facts can risk major breakages for existing users, so we try to do so rarely. We'll keep this in mind the next time we look into how we report hardware information. For now, we're unlikely to change it until the next major redesign.
|Comment by Tommy McNeely [ 2018/03/05 ]|
This should workaround the non-standard hardwaremodel (and OS version for that matter), I think. The test is hardcoded to work against the facts that I submitted, but maybe someone who is better with rspec can make it more generic?
|Comment by Branan Riley [ 2018/05/22 ]|
We're evaluating this as a breaking change for Facter 4. If we decide to take it on, it will be added to FACT-1628