I've tried to reproduce the failure but every time I run it against a daisy in the lab, the test passes :-/
I did notice that the update_engine.log is empty during each of failures.
So my first guess is that calling update_engine_client --update right after calling restart update-engine means the update request is missed
However, on a couple of occasions when I ran it from my desk, even though the test passed the update_engine.log was still empty (empty meaning it just had the first few startup entries, no update requests)
So I am thinking that sometimes when we sign out of chrome (as this test does) update_engine restarts/terminates and the log we collect afterwards is a not the one from the test.
Looking at it again. Daisy is failing before we even sign into Chrome. I'll grab somemore update_engine logs anyway because I think we are not saving the one from the test on daisy.
Comment 1 by dhadd...@chromium.org
, Apr 16 2018