My basic_scaling instance hours are way too high

[Chester Moy]

I have this issue/query posted up on serverfault, but so far, it doesn’t seem
like people think it’s appropriate. So I’m asking here in case my post gets
deleted.

I have a small app hosted on GAE with 2 modules. The default module, frontend is
alotted 28 instance hours. The backend is using basic_scaling: so only has 8
instance hours available.

I noticed that my application has only been running for 2 hours, yet the backend
module has already used up 76% of its quota! This doesn’t make sense to me.
According to the docs, instance hours are calculated by instance uptime (taking
into account multiple instances).

The instance hours should correlate 1:1 with my module’s uptime. I have
max_instances: 1 set. I also checked the running instances just to make sure
that App Engine wasn’t spawning multiple instances. It wasn’t. I get one
instance, but it’s burning up hours faster than it should.

Am I just misinterpreting the docs? If that’s the case, how do I lower the rate
it burns through the quota?

​

[Alex Martelli]

I just posted an answer on serverfault.

TL;DR: it's quite possible that you are misinterpreting the docs, but we can just guess, because you are not giving the crucial information (configuration details, in particular instance class; activity pattern).

It is not the case that "instance hours should correlate 1:1 with [your] module’s uptime", unless your instance class is B1 and your module never goes idle. All larger instance classes are shown in terms of B1 instance-hours, with various multiples (e.g, the default instance class, B2, has a multiple of two: one hour of B2 shows up as two instance-hours). And, each time the module goes idle, billing ceases 15 minutes (elapsed, i.e, wall-clock, time) later.

I believe the docs are clear on this, but if that is not the case then please indicate how you believe things could be made clearer, to help us improve the documentation -- thanks!

Alex

--
You received this message because you are subscribed to the Google Groups "Google App Engine" group.
To unsubscribe from this group and stop receiving emails from it, send an email to google-appengi...@googlegroups.com.
To post to this group, send email to google-a...@googlegroups.com.
Visit this group at http://groups.google.com/group/google-appengine.
To view this discussion on the web visit https://groups.google.com/d/msgid/google-appengine/f0d3c773-b135-4db9-9110-310ce19daeed%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Chester Moy]

Then I did misinterpret the docs. It seemed like pretty explicit language to me,
but I was making some assumptions.

Instance usage is billed by instance uptime, at a given hourly rate.
…
Instances are priced based on an hourly rate determined by the instance class.
…
Manual and basic scaling instances are billed at hourly rates based on uptime.

I took this to mean that instance classes had differing $/hr rates but, what
it actually meant was i_hr/hr. That’s essentially the same thing if one were
already paying for the service. But for someone who’s in the free tier, the wording
was a little confusing.

Thanks for your clarification. The docs are only clear for people already
familiar with cloud computing terminology.

​

[Alex Martelli]

Thanks for clarifying, and, I agree, it would be great to clarify the doc page (I'll be working with the docs team on how exactly to do that -- it's far from obvious, since said page is already big and inevitably complicated).

Meanwhile, could you please accept my answer to your Q on serverfault? This way the issue will show up as "answered" and be findable that way by others who may understandably run into the same issue of unclear communication that you did.

Thanks,

Alex

--

You received this message because you are subscribed to the Google Groups "Google App Engine" group.
To unsubscribe from this group and stop receiving emails from it, send an email to google-appengi...@googlegroups.com.
To post to this group, send email to google-a...@googlegroups.com.
Visit this group at http://groups.google.com/group/google-appengine.

To view this discussion on the web visit https://groups.google.com/d/msgid/google-appengine/9ee9447e-c7d3-4121-b17a-c1c1fb78c021%40googlegroups.com.

[Dimitri Adamou]

What would help is if we could input our set ups (eg basic_scaling / automatic_scaling b/f 1/2/3/4 and have the minimum (no traffic) vs maximum (constant traffic so always up) price ranges

so eg

I'm running

3 B4's on basic_scaling

1 B2 on manual

and front-end F4 scaling with 4 maximum front end (My current monthly is just hitting 1k)

[Alex Martelli]

On Mon, Aug 31, 2015 at 4:00 PM, Dimitri Adamou <dimitri...@formitize.com> wrote:

What would help is if we could input our set ups (eg basic_scaling / automatic_scaling b/f 1/2/3/4 and have the minimum (no traffic) vs maximum (constant traffic so always up) price ranges

so eg

I'm running

3 B4's on basic_scaling

1 B2 on manual

and front-end F4 scaling with 4 maximum front end (My current monthly is just hitting 1k)

Thanks for the suggestion! (And, for once, this feature request is appropriate here, since the cloud pricing calculator does not have its own issue tracker). I've forwarded it to the team that own the cloud pricing calculator, and they will be looking into it.

Alex

 

To view this discussion on the web visit https://groups.google.com/d/msgid/google-appengine/8261ce10-afa6-4d12-8ab5-40a2ce814eaa%40googlegroups.com.