MODULES-7734 Detect installed features with value greater than 1

[GrammatonKlaric]

Add logical OR to registry subkey check for installed features

[glennsarti]

Thanks @GrammatonKlaric for the PR. These numbers look suspiciously like a bit mask. I don't suppose you have any information about how to install SQL Server and get the DWORD of 4.

[GrammatonKlaric]

@glennsarti I do not know how to get the SQL install to do this, but in our organization lately it seems that 1 out of 4 at a minimum will present this error in at least one of the features. I do know that you can manually change the registry to 4 and replicate the exact error I am seeing. You can make this change on any or all of the instance features at "HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Microsoft SQL Server\MSSQL12.MSSQLSERVER\ConfigurationState"

This will reliably create the condition that I am seeing. SQL will be fully functional, but the sqlserver_instance resource will not read the values (and facterdb will show an incorrect features array).

Also, you can correct the condition by simply changing the 4 back to a 1 and then the features array will report correctly and the resource will behave as expected.

[GrammatonKlaric]

I have just discovered a server with values of 2 in the registry (I suspected it would be more than 1 or 4, this is confirmation). Since this registry value is stored as a string, perhaps we should convert to integer and then compare for values greater than 0).
I am conducting research on the server with values of 2 and seeing if the instance is still operational and installed and if the same resolution will work.

[GrammatonKlaric]

I just noticed that this occurred on a server that had an updated version of SQL 2014 installed first, then and older version with older patches were pushed up via hieradata. This would not happen in our production environment, so I'm not sure if the value of 2 is relevant or not.

[GrammatonKlaric]

FYI, I am going to submit a new Pull Request with better code. This is my first contribution to GitHub, so I'm not sure if I should close this PR or not (for comments/history).
Update: I amended the code instead of submitting a new PR. Instead of a logical OR, I have tested converting to integer and doing a numerical comparison. Please review.

[glennsarti]

Sorry @GrammatonKlaric I've been out for a while. You can close this PR but you can still reference it in the comments of the new one.

[GrammatonKlaric]

Instead of making a new PR, I amended this commit with the new code. I figured this was okay since no one is using it yet. Typically I don't code this way, but I am dealing with an operational issue, so the first code was a quick patch, but the new code is ideal and has been tested.

Please use this PR.

[GrammatonKlaric]

@glennsarti Hi Glenn, looks like we lost touch for a while. Is there any reason why this PR cannot be accepted? I have this code running in my organization on hundreds of SQL servers now with no issue.

[glennsarti]

Thanks @GrammatonKlaric Merged.