Fix server state incorrectly reported

[israel-hdez]

Server state was being reported based only on inventory data. This will
report server state based on a combination of inventory data and
availability metrics, if EMS/provider features it.

https://bugzilla.redhat.com/show_bug.cgi?id=1438816

This PR goes together with ManageIQ/manageiq-providers-hawkular#6. Merge order is not important and is not required to merge simultaneously.

When server is running, inventory state is reported:

When server is down, availability metric state is reported:

[israel-hdez]

@rubenvp8510 can you review?

[israel-hdez]

@miq-bot add_label middleware

[dclarizio]

@israel-hdez from what I can tell, the BZ is upstream only, so I labeled as fine/no . . . let me know if this is not the case. Thx, Dan

[abonas]

@israel-hdez from what I can tell, the BZ is upstream only, so I labeled as fine/no . . . let me know if this is not the case. Thx, Dan

@dclarizio why do you think the BZ is upstream only?

[abonas]

@israel-hdez so the "Calculated" is coming from metrics? what if both are non empty? is it correct to take the first one?

[dclarizio]

@abonas

@dclarizio why do you think the BZ is upstream only?

Not sure, I would think any upstream bugs would at least go back to Fine release. Maybe check with the reporter to see if they forgot to set the release field.

[israel-hdez]

so the "Calculated" is coming from metrics? what if both are non empty? is it correct to take the first one?

@abonas Indeed, both values should be non empty most times. The "Calculated" is a combination of the availability metric and the state reported by inventory ("Server State"):

• If availability metric reports "up" value, then "Calculated" will be the same as "Server State" reported by inventory (value will be either "starting", "running" or "restart required")
• Else, then "Calculated" will take the value of the availability metric (either "Down" or "Unknown").

I understand that the UI is provider generic, so I kept both fields. This way, if some other middleware provider doesn't need to "calculate" the server status, it can just fill the "server status". But given that right now, hawkular is the only middleware provider, it will always take the first value. If this is noisy, I can just keep the "calculated" one.

[abonas]

@israel-hdez can you put a short comment in the git commit that will explain what you explained so greatly above? your comment here won't be part of git log, so I'd like the explanation to be captured for whoever looks at the code and the commit why it is written that way.
after that, I hope to see it merged soon cc @dclarizio

[abonas]

@miq-bot add_label fine/yes

[abonas]

@miq-bot remove_label fine/no

[dclarizio]

@israel-hdez @abonas re-ping me when the commit message is changed as this looks ready for merge.

[miq-bot]

Checked commit israel-hdez@29e5626 with ruby 2.2.6, rubocop 0.47.1, and haml-lint 0.20.0
1 file checked, 0 offenses detected
Everything looks good. ⭐

[israel-hdez]

@dclarizio The comment of the git commit is in place. Now, you can merge.

[simaishi]

Fine backport details:

$ git log -1
commit f5e6cc0b2f533fb4da406470312d802f10f8609c
Author: Dan Clarizio <dclarizi@redhat.com>
Date:   Tue Apr 18 02:10:39 2017 -0700

    Merge pull request #987 from israel-hdez/BZ-1438816
    
    Fix server state incorrectly reported
    (cherry picked from commit 9a704488ac200d92f8458fd7e387e8d88ff1a572)