Many unnecessary instance starts in the last few days

[tempy]

Hello all,

I've just been seeing this behavior for the last few days. A request
comes into my lightly-loaded app that already has a nearly-idle
instance running. Instead of using that instance, it starts a new one
and that request suffers the cold-start penalty.

If I look at the "instances" graph on the admin console, then for
these days the blue total line is way over the "billed" and "active"
lines. What could be causing this behavior?

Thanks,
Mike

[C Sowa]

Seconded.  I have a modest, stable little app that runs under the free quota.  The last few days, however, instance hours in the 50 range have been reported on the dashboard, resulting in several hours of downtime due to exceeding quotas.  When I look at the instances chart, I notice that the green (billable) line has been at or very near one all day, and then zero for the out of quota period.

 

Here's where it gets interesting: When I look at billing history for the day, it reports under 24 hours, which is accurately reflects what the instances chart had indicated.  So why am I getting out of quota downtime?  And why is ms / request, normally at 60 or so, skyrocketing to 6000?  There was no corresponding fluctuation in request type or rate.