App Engine app.yaml reference

Region ID

The REGION_ID is an abbreviated code that Google assigns based on the region you select when you create your app. The code does not correspond to a country or province, even though some region IDs may appear similar to commonly used country and province codes. For apps created after February 2020, REGION_ID.r is included in App Engine URLs. For existing apps created before this date, the region ID is optional in the URL.

Learn more about region IDs.

You configure your App Engine app's settings in the app.yaml file. This file specifies how URL paths correspond to request handlers and static files. The app.yaml file also contains information about your app's code, such as the runtime and the latest version identifier.

Each service in your app has its own app.yaml file, which acts as a descriptor for its deployment. You must first create the app.yaml file for the default service before you can create and deploy app.yaml files for additional services within your app.

Directory structure

To learn more about structuring multiple services in your app, see Structuring Web Services in App Engine.

Example

The following is an example of an app.yaml file for a Python 2 application:

runtime: python27
api_version: 1
threadsafe: true

handlers:
- url: /
  script: home.app

- url: /index\.html
  script: home.app

- url: /stylesheets
  static_dir: stylesheets

- url: /(.*\.(gif|png|jpg))$
  static_files: static/\1
  upload: static/.*\.(gif|png|jpg)$

- url: /admin/.*
  script: admin.app
  login: admin

- url: /.*
  script: not_found.app

A script: directive can contain either a file path ending in .py, which means the script uses CGI, or a Python module path, with package names separated by dots, which means the script uses WSGI.

Syntax

The syntax of app.yaml is the YAML format.

The YAML format supports comments. A line that begins with a pound (#) character is ignored:

# This is a comment.

URL and file path patterns use POSIX extended regular expression syntax, excluding collating elements and collation classes. Back-references to grouped matches (e.g. \1) are supported, as are these Perl extensions: \w \W \s \S \d \D.

Runtime and app elements

Element Description
application

The recommended approach is to remove the application element from your app.yaml file and instead, use a command-line flag to specify your application ID:

  • To use the gcloud app deploy command, you must specify the --project flag:
    gcloud app deploy --project [YOUR_PROJECT_ID]

For more information about using these commands, see Deploying Your App.

The application ID is the Google Cloud console project ID that you specified when you created the application in the Google Cloud console.

api_version

Required. The version of the API in the given runtime environment that is used by your app.

This field is deprecated in newer App Engine runtimes.

When Google announces support for a new version of a runtime environment's API, your deployed app will continue to use the one for which it was written. To upgrade your app to a new version of the API, you change this value and then redeploy your app to App Engine. When you specify the 1 value, the latest supported runtime environment is used each time you deploy that app (currently ).

At this time, App Engine has one version of the python27 runtime environment: 1

auto_id_policy Optional. If you are setting entity identifiers automatically, you can change the method employed by setting the auto ID policy. The following are valid options:
default
Default. Uses scattered auto IDs that are large well-distributed integers that are small enough to be represented by 64-bit floats.
legacy
The legacy option will be deprecated in a future release and will eventually be removed. For more information, see the blog post announcing this change.
builtins

Optional. The Python 2 SDK includes a number of built-in handlers for common application functions. The builtins directive allows you to include specific handlers in app.yaml.

This field is deprecated in the Python 3 runtime.

The following built-in handlers are available for your use:

appstats
Enables Appstats at /_ah/stats/, which you can use to measure your application's performance. In order to use Appstats, you also need to install the event recorder.
deferred
Enables the deferred handler at /_ah/queue/deferred. This builtin allows developers to use deferred.defer() to simplify the creation of Task Queue tasks.
remote_api
Enables the remote_api builtin at /_ah/remote_api/. This builtin allows remote applications with the proper credentials to access the datastore remotely.
Example:
builtins:
- deferred: on
- appstats: on

The builtins directive is a special instance of the includes directive. Each builtin directive is equivalent, in Python, to an includes directive with an expanded path. For example:

builtins:
- name: on

Is equivalent to:

includes:
- $PYTHON_LIB/google/appengine/ext/builtins/name/

When you use builtins in your app.yaml file, any handlers that are defined by in the built-in include.yaml file will supersede any handlers that you define in your app.yaml file. However, if you include a file that then uses builtins or includes, the handlers are added by order of the include hierarchy. In other words, the handlers of the "parent" include are added before the builtins of the "child" includes, and so on.

For example, consider the following app.yaml, which uses the built-in appstats handlers:

handlers:
- url: /.*
  script: main.app
builtins:
- appstats: on

The resulting list of handlers is:

[/_ah/stats, /.*]

If the app.yaml uses an includes directive:

includes:
- included.yaml

And the included.yaml file uses builtins:

handlers:
- url: /.*
  script: main.app
builtins:
- appstats: on

The resultant list of handlers is now:

[/.*, /_ah/stats]

The order of placement of the builtins clause in a .yaml file doesn't change the behavior.

default_expiration

Optional. Sets a global default cache period for all static file handlers for an application. You can also configure a cache duration for specific static file handlers. The value is a string of numbers and units, separated by spaces, where units can be d for days, h for hours, m for minutes, and s for seconds. For example, "4d 5h" sets cache expiration to 4 days and 5 hours after the file is first requested. If omitted, the production server sets the expiration to 10 minutes.

Example:
runtime: python27
api_version: 1
threadsafe: true

default_expiration: "4d 5h"

handlers:
# ...

For more information, see Cache expiration.

env_variables

Optional. You can define environment variables in your app.yaml file to make them available to your app. Ensure that the key in Environment variable(s) matches the expression '[a-zA-Z_][a-zA-Z0-9_]*' (start with alphabet or "_" followed by any alphanumeric or "_").

Environment variables that are prefixed with GAE are reserved for system use and not allowed in the app.yaml file.

These variables will be available in the os.environ dictionary:
env_variables:
  DJANGO_SETTINGS_MODULE: "myapp.settings"
error_handlers

Optional. Used to configure custom error pages that are returned for different error types.

This element can contain the following elements:

error_code
Optional. The error_code can be one of the following:
over_quota
Indicates the app has exceeded a resource quota
timeout
Served if a deadline is reached before there is a response from your app.

The error_code is optional; if it's not specified, the given file is the default error response for your app.

file
Each file entry indicates a static file that should be served in place of the generic error response. If you specify a file element without a corresponding error_code element, the static file will be the default error page for your app. The custom error data must be less than 10 kilobytes.
Example
error_handlers:
  - file: default_error.html

  - error_code: over_quota
    file: over_quota.html
handlers

Required. A list of URL patterns and descriptions of how they should be handled. App Engine can handle URLs by executing application code, or by serving static files uploaded with the code, such as images, CSS, or JavaScript.

See the Handlers and sub-elements syntax

includes

Optional. The includes directive allows you to include the configuration file for any library or service throughout your application. For example, you might include a user administration library as follows:

includes:
- lib/user_admin.yaml

App Engine resolves the included path in the following order:

  • Absolute or relative path to the working directory. The specified path resolves to a file.
  • Relative to the application's directory, which is also known as the basepath. The basepath and path resolve to a file.
  • Relative to the file that included the current file. The location of the referring file and the include path resolve to the included file.

If the include directive specifies a directory, then App Engine looks in that directory for a file called include.yaml. If the include directive is a file, then that specific file is included. Using includes retrieves only the following types of directives from the destination file (if present):

Included skip_files patterns are added to those in the including app.yaml, or to the default list if there is no explicit list in app.yaml. Note that skip_files compares absolute paths.

inbound_services

Optional. Applications must enable those services before it can receive inbound requests. You can enable the service for a Python 2 app by including an inbound_services section in the app.yaml file.

The following inbound services are available:

mail
Allows your application to receive mail.
warmup
Enables warmup requests. See Configuring Warmup Requests.
Example:
inbound_services:
  - mail
  - warmup
instance_class

Optional. The instance class for this service.

The following values are available depending on your service's scaling:

Automatic scaling
F1, F2, F4, F4_1G
Default: F1

Optionally use the automatic_scaling element to change default settings for automatic scaling, such as minimum and maximum number of instances, latency, and concurrent connections.

Note: If instance_class is set to F2 or higher, you can optimize your instances by setting max_concurrent_requests to a value higher than the default value of 10. To determine the optimal value, gradually increase it and monitor the performance of your application.

Basic and manual scaling
B1, B2, B4, B4_1G, B8
Default: B2

Basic and manual instance classes require you to specify either the basic_scaling element or the manual_scaling element.

libraries

Optional. The Python 2.7 runtime includes some third-party libraries. Some of these are available by default; others are only available if configured. You can specify which version you want to use by specifying the name and version values.

This field is deprecated in the Python 3 runtime.

libraries:
- name: PIL
  version: "1.1.7"
- name: webob
  version: "latest"
        

Note than when you specify latest, the SDK determines the latest library version at deployment time. Once deployed, the library version will not change. The only way to get a different version of the library is to deploy again.

If you're developing an application that doesn't have users yet: you don't need to track new versions. But if your application is being actively used, beware: you might be surprised that your application starts using a new not-backward-compatible library version.

For a list of the included third-party libraries, see Third-party Libraries. You can use additional pure-python third-party libraries by installing them into a local directory.

If you are using the flexible environment, see Using Python libraries in the flexible environment.

module

Note: Modules are now named Services.

To manage your app with the gcloud CLI, use the service element instead.

runtime

Required. The name of the runtime environment that is used by your app. For example, to specify Python 2.7, use:

runtime: python27
service

Services were formerly known as Modules.

Supported only by the gcloud CLI or gcloud CLI-based plugins, for example: gcloud app deploy .

Required if creating a service. Optional for the default service. Each service and each version must have a name. A name can contain numbers, letters, and hyphens. The combined length of VERSION-dot-SERVICE-dot-PROJECT_ID, where VERSION is the name of your version, SERVICE is the name of your service, and PROJECT_ID is your project ID, cannot be longer than 63 characters and cannot start or end with a hyphen. Choose a unique name for each service and each version. Don't reuse names between services and versions.

Example:
service: service-name

Note: The gcloud app deploy command is backwards compatible and supports existing app.yaml files that include services declared as modules, for example:

module: service-name
service_account

Optional. The service_account element lets you specify a user-managed service account as the identity for the version. The specified service account is used when accessing other Google Cloud services and executing tasks.

Example:
service_account: [SERVICE_ACCOUNT_NAME]@[PROJECT_ID].iam.gserviceaccount.com
skip_files

Optional. The skip_files element specifies which files in the application directory are not to be uploaded to App Engine. The value is either a regular expression, or a list of regular expressions. Any filename that matches any of the regular expressions is omitted from the list of files to upload when the application is uploaded. Filenames are relative to the project directory.

The skip_files has the following default:

skip_files:
- ^(.*/)?#.*#$
- ^(.*/)?.*~$
- ^(.*/)?.*\.py[co]$
- ^(.*/)?.*/RCS/.*$
- ^(.*/)?\..*$

The default pattern excludes Emacs backup files with names of the form #...# and ...~, .pyc and .pyo files, files in an RCS revision control directory, and Unix hidden files with names beginning with a dot (.).

To extend the above regular expression list, copy and paste the above list into your app.yaml and add your own regular expressions. For example, to skip files whose names end in .bak in addition to the default patterns, add an entry like this for skip_files:

skip_files:
- ^(.*/)?#.*#$
- ^(.*/)?.*~$
- ^(.*/)?.*\.py[co]$
- ^(.*/)?.*/RCS/.*$
- ^(.*/)?\..*$
- ^(.*/)?.*\.bak$

To skip a full directory, add the directory name to the list. For example, to skip a directory named logs, add the following line to the previously described ones:

skip_files:
- logs/
threadsafe

Required. Configures your application to use concurrent requests. If using Python's threading library, the thread-local data, as returned by threading.local(), is cleared after each request.

This field is deprecated in the Python 3 runtime.

threadsafe: [true | false]

Note: The threadsafe directive is required for Python 2.7 applications. threadsafe: true requires that all script handlers be WSGI ones. That is, each script must be specified in a script: directive a using Python module path, with package names separated by dots. The last component of a script: directive using a Python module path is the name of a global variable in the service: that variable must be a WSGI app, and is usually called app by convention.

version

The recommended approach is to remove the version element from your app.yaml file and instead, use a command-line flag to specify your version ID:

  • To use the gcloud app deploy command, you specify the -v flag:
    gcloud app deploy -v [YOUR_VERSION_ID]

For more information about using this command, see Deploying Your App.

An identifier for the version of your application code that you deploy to App Engine.

The version ID can contain lowercase letters, digits, and hyphens. It cannot begin with the prefix ah- and the names default and latest are reserved and cannot be used.

Note: Version names should begin with a letter, to distinguish them from numeric instances which are always specified by a number. This avoids the ambiguity with URLs like 123-dot-my-service.uc.r.appspot.com, which can be interpreted two ways: If version "123" exists, the target will be version "123" of the given service. If that version does not exist, the target will be instance number 123 of the default version of the service.

Each version of an application retains its own copy of app.yaml. When an application is uploaded, the version mentioned in the app.yaml file being uploaded is the version that gets created or replaced by the upload. An administrator can change which version of the application is serving traffic by using the Google Cloud console, and can also test other versions before configuring them to receive traffic.

vpc_access_connector

Optional. Configures your application to use a Serverless VPC Access connector, enabling the application to send requests to internal resources in your VPC network. For more information, see Connecting to a VPC network.

name
String literal. Specify the fully-qualified name of your Serverless VPC Access connector in quotes:
"projects/PROJECT_ID/locations/REGION/connectors/CONNECTOR_NAME"
egress_setting
Optional. Default is private-ranges-only. The egress_setting can be one of the following:
private-ranges-only
Default. Requests to internal IP addresses are sent through the Serverless VPC Access connector into the connected VPC network. Requests to external IP addresses are sent to the public internet.
all-traffic
All requests are sent through the Serverless VPC Access connector into the connected VPC network.
Example
vpc_access_connector:
  name: "projects/PROJECT_ID/locations/REGION/connectors/CONNECTOR_NAME"
  egress_setting: all-traffic

Handlers element

The handlers element is a required element in the app.yaml configuration file. The element provides a list of URL patterns and descriptions of how they should be handled. App Engine can handle URLs by executing application code, or by serving static files uploaded with the code, such as images, CSS, or JavaScript.

Patterns are evaluated in the order they appear in the app.yaml file, from top to bottom. The first mapping whose pattern matches the URL is the one used to handle the request.

The following table lists the subelements of the handlers element that control the behavior for scripts, static files, static directories, and other settings.

Element Description
application_readable Optional. Boolean. By default, files declared in static file handlers are uploaded as static data and are only served to end users. They cannot be read by an application. If this field is set to true, the files are also uploaded as code data so your application can read them. Both uploads are charged against your code and static data storage resource quotas.

This field is deprecated in newer App Engine runtimes.

expiration Optional. The length of time a static file served by this handler should be cached by web proxies and browsers. The value is a string of numbers and units, separated by spaces, where units can be d for days, h for hours, m for minutes, and s for seconds. For example, "4d 5h" sets cache expiration to 4 days and 5 hours after the file is first requested. If omitted, the application's default_expiration is used. See Cache expiration for more details.
http_headers

Optional. You can set HTTP headers for responses of your static file or directory handlers. If you need to set HTTP headers in your script handlers, you should instead do that in your app's code. For information about which response headers influence caching, see Caching static content.

Example
handlers:
- url: /images
  static_dir: static/images
  http_headers:
    X-Foo-Header: foo
    X-Bar-Header: bar value
    vary: Accept-Encoding
  # ...

CORS Support

One important use of this feature is to support cross-origin resource sharing (CORS), such as accessing files hosted by another App Engine app.

For example, you could have a game app mygame.uc.r.appspot.com that accesses assets hosted by myassets.uc.r.appspot.com. However, if mygame attempts to make a JavaScript XMLHttpRequest to myassets, it will not succeed unless the handler for myassets returns an Access-Control-Allow-Origin: response header containing the value http://mygame.uc.r.appspot.com.

Here is how you would make your static file handler return that required response header value:

handlers:
- url: /images
  static_dir: static/images
  http_headers:
    Access-Control-Allow-Origin: https://mygame.uc.r.appspot.com
  # ...

Note: if you wanted to allow everyone to access your assets, you could use the wildcard '*', instead of https://mygame.uc.r.appspot.com.

mime_type

Optional. If specified, all files served by this handler will be served using the specified MIME type. If not specified, the MIME type for a file will be derived from the file's filename extension. If the same file is uploaded with multiple extensions, the resulting extension can depend on the order in which the uploads occurred.

For more information about the possible MIME media types, see the IANA MIME Media Types website

redirect_http_response_code

Optional. redirect_http_response_code is used with the secure setting to set the HTTP response code returned when performing a redirect required by how the secure setting is configured. redirect_http_response_code element has the following possible values:

301
Moved Permanently response code.
302
Found response code.
303
See Other response code.
307
Temporary Redirect response code.
Example
handlers:
- url: /youraccount/.*
  script: accounts.app
  login: required
  secure: always
  redirect_http_response_code: 301

When a user's request is redirected, the HTTP status code will be set to the value of the redirect_http_response_code parameter. If the parameter is not present, 302 will be returned.

script

Optional. Specifies the path to the script from the application root directory:

handlers:
# The root URL (/) is handled by the WSGI application named
# "app" in home.py. No other URLs match this pattern.
- url: /
  script: home.app

# The URL /index.html is also handled by the home.py script.
- url: /index\.html
  script: home.app

# A regular expression can map parts of the URL to the
# path of the script.
- url: /browse/(books|videos|tools)
  script: \1.catalog.app

# All other URLs use the WSGI application named in "app"
# in not_found.py.
- url: /.*
  script: not_found.app

A script: directive must be a python import path, for example, package.module.app that points to a WSGI application. The last component of a script: directive using a Python module path is the name of a global variable in the module: that variable must be a WSGI app, and is usually called app by convention.

Note: just like for a Python import statement, each subdirectory that is a package must contain a file named __init__.py.

In newer App Engine runtimes, the the behavior of this field has changed.

secure Optional. Any URL handler can use the secure setting, including script handlers and static file handlers. The secure element has the following possible values:
optional
Both HTTP and HTTPS requests with URLs that match the handler succeed without redirects. The application can examine the request to determine which protocol was used, and respond accordingly. This is the default when secure is not provided for a handler.
never
Requests for a URL that match this handler that use HTTPS are automatically redirected to the HTTP equivalent URL. When a user's HTTPS request is redirected to be an HTTP request, the query parameters are removed from the request. This prevents a user from accidentally submitting query data over a non-secure connection that was intended for a secure connection.
always
Requests for a URL that match this handler that do not use HTTPS are automatically redirected to the HTTPS URL with the same path. Query parameters are preserved for the redirect.
Example
handlers:
- url: /youraccount/.*
  script: accounts.app
  login: required
  secure: always

The development web server does not support HTTPS connections. It ignores the secure parameter, so paths intended for use with HTTPS can be tested using regular HTTP connections to the development web server.

To target a specific version of your app using the REGION_ID.r.appspot.com domain, you replace the periods that would usually separate the subdomain components of the URL with the string "-dot-", for example:
https://VERSION_ID-dot-default-dot-PROJECT_ID.REGION_ID.r.appspot.com

To use custom domains with HTTPS, you must first activate and configure SSL certificates for that domain.

Google Accounts sign-in and sign-out are always performed using a secure connection, unrelated to how the application's URLs are configured.

static_dir

Optional. The path to the directory containing the static files, from the application root directory. Everything after the end of the matched url pattern is appended to static_dir to form the full path to the requested file.

Each file in the static directory is served using the MIME type that corresponds with its filename extension unless overridden by the directory's mime_type setting. All of the files in the given directory are uploaded as static files, and none of them can be run as scripts.

All files in this directory are uploaded with your app as static files. App Engine stores and serves static files separately from your app's files. Static files are not available in the app's file system by default. This can be changed by setting the application_readable option to true.

Example:
handlers:
# All URLs beginning with /stylesheets are treated as paths to
# static files in the stylesheets/ directory.
- url: /stylesheets
  static_dir: stylesheets
  # ...
static_files

Optional. A static file pattern handler associates a URL pattern with paths to static files uploaded with the application. The URL pattern regular expression can define regular expression groupings to be used in the construction of the file path. You can use this instead of static_dir to map to specific files in a directory structure without mapping the entire directory.

Example:
handlers:
# All URLs ending in .gif .png or .jpg are treated as paths to
# static files in the static/ directory. The URL pattern is a
# regular expression, with a grouping that is inserted into the
# path to the file.
- url: /(.*\.(gif|png|jpg))$
  static_files: static/\1
  upload: static/.*\.(gif|png|jpg)$
  # ...

App Engine stores and serves static files separately from application files. Static files are not available in the application's file system by default. This can be changed by setting the application_readable option to true.

Static files cannot be the same as application code files. If a static file path matches a path to a script used in a dynamic handler, the script will not be available to the dynamic handler.

upload

Optional. A regular expression that matches the file paths for all files that will be referenced by this handler. This is necessary because the handler cannot determine which files in your application directory correspond with the given url and static_files patterns. Static files are uploaded and handled separately from application files. The example above might use the following upload pattern: archives/(.*)/items/(.*)

url

Required element under handlers. The URL pattern, as a regular expression. The expression can contain groupings that can be referred to in the file path to the script with regular expression back-references. For example, /profile/(.*)/(.*) would match the URL /profile/edit/manager and use edit and manager as the first and second groupings.

The URL pattern has some differences in behavior when used with the following elements:

static_dir
Uses a URL prefix. The regular express pattern should not contain groupings when used with the static_dir element. All URLs that begin with this prefix are handled by this handler, using the portion of the URL after the prefix as part of the file path.
static_files
A static file pattern handler associates a URL pattern with paths to static files uploaded with the application. The URL pattern regular expression can define regular expression groupings to be used in the construction of the file path. You can use this instead of static_dir to map to specific files in a directory structure without mapping the entire directory.

Scaling elements

The elements in following table configure how your application scales. To learn more about how App Engine apps scale, see Scaling types.

Element Description
automatic_scaling

Optional. Applicable only for applications that use an instance class of F1 or higher.

Specify this element to change default settings for automatic scaling, such as setting minimum and maximum levels for number of instances, latency, and concurrent connections for a service.

This element can contain the following elements:

max_instances
Optional. Specify a value between 0 and 2147483647, where zero disables the setting.

This parameter specifies the maximum number of instances for App Engine to create for this module version. This is useful to limit the costs of a module.

min_instances
Optional. The minimum number of instances for App Engine to create for this module version. These instances serve traffic when requests arrive, and continue to serve traffic even when additional instances are started up as required to handle traffic.

Specify a value from 0 to 1000. You can set the parameter to the value 0 to allow scaling to 0 instances to lower costs when no requests are being served. Note that you are charged for the number of instances specified whether they are receiving traffic or not.

max_idle_instances

Optional. The maximum number of idle instances that App Engine should maintain for this version. Specify a value from 1 to 1000. If not specified, the default value is automatic, which means App Engine will manage the number of idle instances. Keep the following in mind:

  • A high maximum reduces the number of idle instances more gradually when load levels return to normal after a spike. This helps your application maintain steady performance through fluctuations in request load, but also raises the number of idle instances (and consequent running costs) during such periods of heavy load.
  • A low maximum keeps running costs lower, but can degrade performance in the face of volatile load levels.

Note: When settling back to normal levels after a load spike, the number of idle instances can temporarily exceed your specified maximum. However, you will not be charged for more instances than the maximum number you've specified.

min_idle_instances

Optional: The number of additional instances to be kept running and ready to serve traffic for this version.

App Engine calculates the number of instances necessary to serve your current application traffic based on scaling settings such as target_cpu_utilization and target_throughput_utilization. Setting min_idle_instances specifies the number of instances to run in addition to this calculated number. For example, if App Engine calculates that 5 instances are necessary to serve traffic, and min_idle_instances is set to 2, App Engine will run 7 instances (5, calculated based on traffic, plus 2 additional per min_idle_instances).

Note that you are charged for the number of instances specified whether they are receiving traffic or not. Keep the following in mind:

  • A low minimum helps keep your running costs down during idle periods, but means that fewer instances might be immediately available to respond to a sudden load spike.
  • A high minimum allows you to prime the application for rapid spikes in request load. App Engine keeps the minimum number of instances running to serve incoming requests. You are charged for the number of instances specified, whether or not they are handling requests.

    If you set a minimum number of idle instances, pending latency will have less effect on your application's performance.

target_cpu_utilization
Optional. Specify a value between 0.5 and 0.95. The default is 0.6.

This parameter specifies the CPU usage threshold at which new instances will be started to handle traffic, enabling you to balance between performance and cost, with lower values increasing performance and increasing cost, and higher values decreasing performance but also decreasing cost. For example, a value of 0.7 means that new instances will be started after CPU usage reaches 70 percent.

target_throughput_utilization
Optional. Specify a value from 0.5 to 0.95. The default is 0.6.

Used with max_concurrent_requests to specify when a new instance is started due to concurrent requests. When the number of concurrent requests reaches a value equal to max_concurrent_requests times target_throughput_utilization, the scheduler tries to start a new instance.

max_concurrent_requests

Optional. The number of concurrent requests an automatic scaling instance can accept before the scheduler spawns a new instance (Default: 10, Maximum: 1000).

Used with target_throughput_utilization to specify when a new instance is started due to concurrent requests. When the number of concurrent requests reaches a value equal to max_concurrent_requests times target_throughput_utilization, the scheduler tries to start a new instance.

We recommend you do not set max_concurrent_requests to less than 10 unless you need single threading. A value of less than 10 is likely to result in more instances being created than you need for a threadsafe app, and that may lead to unnecessary cost.

If this setting is too high, you might experience increased API latency. Note that the scheduler might spawn a new instance before the actual maximum number of requests is reached.

max_pending_latency

The maximum amount of time that App Engine should allow a request to wait in the pending queue before starting additional instances to handle requests so that pending latency is reduced. When this threshold is reached, it is a signal to scale up, and results in an increase in the number of instances. If not specified, the default value is automatic. This means requests can remain in the pending queue for up to 10s, the maximum pending request time limit, before new instance starts are triggered.

A low maximum means App Engine will start new instances sooner for pending requests, improving performance but raising running costs.

A high maximum means users might wait longer for their requests to be served (if there are pending requests and no idle instances to serve them), but your application will cost less to run.

min_pending_latency

An optional element you can set to specify the minimum amount of time that App Engine should allow a request to wait in the pending queue before starting a new instance to handle it. Specifying a value can lower running costs but increase the time users must wait for their requests to be served.

For free apps, the default value is 500ms. For paid apps, the default value is 0.

This element works together with the max_pending_latency element to determine when App Engine creates new instances. If pending requests are in the queue:

  • Less than the min_pending_latency you specify, App Engine will not create new instances.
  • More than max_pending_latency, App Engine will try to create a new instance.
  • Between the time specified by min_pending_latency and max_pending_latency, App Engine will try to reuse an existing instance. If no instances are able to process the request before max_pending_latency, App Engine will create a new instance.
Example
automatic_scaling:
  target_cpu_utilization: 0.65
  min_instances: 5
  max_instances: 100
  min_pending_latency: 30ms
  max_pending_latency: automatic
  max_concurrent_requests: 50
basic_scaling

Applications that use an instance class of B1 or higher must specify either this element or manual_scaling.

This element enables basic scaling of instance classes B1 and higher, can contain the following elements:

max_instances
Required. The maximum number of instances for App Engine to create for this service version. This is useful to limit the costs of a service.
idle_timeout
Optional. The instance will be shut down this amount of time after receiving its last request. The default is 5 minutes (5m).
Example
basic_scaling:
  max_instances: 11
  idle_timeout: 10m
manual_scaling

Applications that use an instance class of B1 or higher must specify either this element or basic_scaling.

This element enables manual scaling of instance classes B1 and higher, and can contain the following element:

instances
The number of instances to assign to the service at the start.
Example
manual_scaling:
  instances: 5