856891 - update the smoketests to account for dead machines

Status: RESOLVED WONTFIX

(Testing :: General, defect)

Description

[Joel Maher ( :jmaher ) (UTC -8)]

While working on the smoketests, I have always been required to have a machine that is up and ready to test, not a machine in any given state.

To make things more robust and reliable, I have added a reboot at the beginning of each smoketest cycle.  This isn't that simple because I have needed to hack up devicemanagerSUT to not get the deviceroot and attempt connections to the device during initialization.

Comment 1

[Joel Maher ( :jmaher ) (UTC -8)]

Attachment: fix devicemanager to not do anything on initialization (1.0)

Comment 2

[William Lachance (:wlach)]

Comment on attachment 732144 [details] [diff] [review]
fix devicemanager to not do anything on initialization (1.0)

As discussed on irc, I think it might be preferable to call reboot_from_relay directly from sut_tools, since that more explicitly maps to what we're trying to do here and doesn't require adding stuff to the devicemanager constructor.

r- for now, but I could be missing something. I could change my mind given a compelling argument.

Comment 3

[Joel Maher ( :jmaher ) (UTC -8)]

I finally got around to looking at this again, and here is what needs to be done:
1) reboot the device
  * fairly easy: https://hg.mozilla.org/build/tools/file/tip/sut_tools/sut_lib.py#l566
2) ensure it comes online
3) start the test

#2 requires us to connect via devicemanager.  I could rewrite the connection logic from devicemanager inside of sut_lib or the smoketest.  Also while testing on a local panda without a relay to trigger a reboot #1 would fail without devicemanager.

I am open to more suggestions...

Comment 4

[Mark Côté [:mcote]]

Deprecating Testing :: Infrastructure.

Comment 5

[Joel Maher ( :jmaher ) (UTC -8)]

sutagent is not used, devicemanagerSUT is only used for negatus, not for devices anymore.