Re: pcp updates: changes for a QA world that avoids systemd

On Fri, Feb 10, 2017 at 7:33 AM, Ken McDonell <[email protected]> wrote:

> These tests were passing on systems without systemd, but there were
>     assorted failures when systemd was active, mostly as a result of
>     the test stopping a PCP daemon and then the test and systemd getting
>     in an arm wrestle over who's restarting the daemon ... the QA tests
>     invariably lose these battles.
>

Ken, perhaps we could introduce a pcpqa.service and then force the restart
ordering with some 'after', 'before' and 'wants' rules in the various other
pcp services? Not exactly sure how it would work but it does provide a way
to enforce service precedence and dependencies. This might lead to a more
deterministic QA environment? Maybe we could work something out - seems to
me now that systemd has more or less taken over the RC universe (despite
many complaints) we should be QA'ing with it. Thoughts?