Move show_console to the client library

[xuhang57]

Simply move show_console to the client library.

Relates to: https://github.com/CCI-MOC/hil/issues/924

[coveralls]

Pull Request Test Coverage Report for Build 1713

• 9 of 11 (81.82%) changed or added relevant lines in 3 files are covered.
• 2 unchanged lines in 1 file lost coverage.
• Overall coverage increased (+0.2%) to 59.576%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
hil/cli.py	0	1	0.0%
hil/ext/obm/mock.py	6	7	85.71%
<!--	Total:	9	11	81.82%	-->

Files with Coverage Reduction	New Missed Lines	%
hil/ext/obm/ipmi.py	2	53.85%
<!--	Total:	2		-->

Totals
Change from base Build 1686:	0.2%
Covered Lines:	2109
Relevant Lines:	3540

---

💛 - Coveralls

[naved001]

just some pep8 failures. You also might wanna add a unit test for client library.

[xuhang57]

Hmm, I am still getting a

FailedAPICallException: The console log for node-01 does not exist.

I started the console before I show it. Did I miss anything here?

[zenhack]

If it's using the mock obm driver, it looks like actually showing the console isn't implemented (hil/ext/obmd/mock.py). If not for the fact that we're looking at migrating stuff to obmd soonish, I might think it was worth stubbing more of this out and maybe putting some logic in there to just e.g. set a boolean when start/stop console are called, and check it in get_console, returning ether None or a string depending. That may actually not be a whole lot of work, in which case it might make sense to do it anyway, but it also seems sensible to just say screw it, and revisit the coverage of the console related calls when we re-do that API.

[naved001]

I am happy with this PR, this seems simple enough to me. We can either update the mock driver to handle get_console or revisit it later (meaning not have the client lib test for it). I am fine either ways. Alternatively, can we use the ipmi driver with dry_run set to true?

[xuhang57]

Thanks, guys!

[naved001]

Turns out the client library was using the ipmi driver rather than the mock driver. Any reason? I changed it to use the mock driver and put some logic like you suggested @zenhack

[naved001]

On a slightly related note, I don't understand why the client library is registering all sorts of drivers when it's not using those. What's the intention here?

[naved001]

@zenhack done.

I'll let @xuhang57 merge master into this (or rebase) after #978 is in.

[zenhack]

I'm happy.