Max number of concurrent requests in manual scaling

[Jungho Ahn]

Hello,

I'm trying to do load tests on my app engine but I couldn't get more than about 8 qps.

Each request has long latency w/ external accesses and the cpu utilization is very low.

Is there any limitation on the max active requests? The app is running in manual scaling.

Thanks,

[Adam (Cloud Platform Support)]

What runtime are you using? Are you using the Standard or Flexible environment? Could you post some details from your app.yaml / appengine-web.xml?

[Jungho Ahn]

We're running on Flexible. Here is my app.yaml:

service: ranking

runtime: python

runtime_config:

    python_version: 2

vm: true

entrypoint: gunicorn -b :$PORT main:app

resources:

  cpu: 32

  memory_gb: 120

manual_scaling:

  instances: 3

We're testing various configurations of gunicorn like

gunicorn -b :$PORT --threads=64 main:app.

Until now we could get QPS around 28 with CPU utilization 70% and we're trying to push more.

I'm just wondering if there is any hard limit on in-flight requests.

--
You received this message because you are subscribed to a topic in the Google Groups "Google App Engine" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/google-appengine/xzVWKR-ulQ0/unsubscribe.
To unsubscribe from this group and all its topics, send an email to google-appengine+unsubscribe@googlegroups.com.
To post to this group, send email to google-appengine@googlegroups.com.
Visit this group at https://groups.google.com/group/google-appengine.
To view this discussion on the web visit https://groups.google.com/d/msgid/google-appengine/d5f7bdb4-177f-4c7c-8624-4282de3731fa%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

Jungho Ahn

[Adam (Cloud Platform Support)]

By default you're only running gunicorn executing one serial request at a time. At a minimum you should specify the number of workers eg.

entrypoint: gunicorn -b :$PORT -w 8 main:app

There doesn't appear to be a '--threads' option to gunicorn. See the docs for more information about tuning the number of workers and setting the worker class. The FAQ recommends 2-4 workers per virtual core.

On Friday, August 5, 2016 at 4:41:21 PM UTC-4, Jungho Ahn wrote:

We're running on Flexible. Here is my app.yaml:

service: ranking

runtime: python

runtime_config:

    python_version: 2

vm: true

entrypoint: gunicorn -b :$PORT main:app

resources:

  cpu: 32

  memory_gb: 120

manual_scaling:

  instances: 3

We're testing various configurations of gunicorn like

gunicorn -b :$PORT --threads=64 main:app.

Until now we could get QPS around 28 with CPU utilization 70% and we're trying to push more.

I'm just wondering if there is any hard limit on in-flight requests.