Azure Function is not creating the new resources and environment for each request when using the docker container

Question

Azure Function is not creating the new resources and environment for each request when using the docker container

Elizaveta Omelyusik 26

When deploying a docker container to Azure Functions it is supposed that azure will create new resources on each request so that the environment for each request will not overlap.
But it turns out that each request is processed by the same container.
Could you please clarify how to use the Azure Function with the independent environments (so that each request is handled in its own container)?

Pramod Valavala 20,656 Reputation points Microsoft Employee Moderator

2021-06-23T11:10:28.667+00:00

@Elizaveta Omelyusik When deploying a new container to a function app, it would replace the existing one. Are you looking to running both the containers simultaneously?
Elizaveta Omelyusik 26 Reputation points

2021-06-24T06:17:21.743+00:00

@Pramod Valavala No, the issue is in the request handling. Azure function does not create container each time the request comes. it creates one container and it handles every incoming request in that container. I was wondering how to deal with it

Answer accepted by question author

0 additional answers

Your answer

Pramod Valavala 20,656 Reputation points Microsoft Employee Moderator

2021-06-23T11:10:28.667+00:00

@Elizaveta Omelyusik When deploying a new container to a function app, it would replace the existing one. Are you looking to running both the containers simultaneously?
Elizaveta Omelyusik 26 Reputation points

2021-06-24T06:17:21.743+00:00

@Pramod Valavala No, the issue is in the request handling. Azure function does not create container each time the request comes. it creates one container and it handles every incoming request in that container. I was wondering how to deal with it

Answer 1

Pramod Valavala 20,656 Microsoft Employee Moderator

@Elizaveta Omelyusik This is the expected behavior since a single container can handle many requests and would eventually scale out as many more requests come in at the same time. Or if each request is compute intensive, then that could trigger a scale out as well.

If your scenario demands lower concurrency, you could tweak the host.json settings for HTTP to reduce the number of requests handled per instance, forcing it to scale out.

Elizaveta Omelyusik 26 Reputation points

2021-06-25T12:03:11.46+00:00

@Pramod Valavala Yes, each request is quite consuming. I tried to set "maxConcurrentRequests" to 1, but it doesn't help. Functions still executed in parallel and it affects their results.
Pramod Valavala 20,656 Reputation points Microsoft Employee Moderator

2021-06-28T10:15:12.34+00:00

@Elizaveta Omelyusik With this setting, each instance will just handle one request at a time but would cause the function app to scale causing multiple invocations across instances.

You could limit the scale out of the function app in addition to the above setting.
Elizaveta Omelyusik 26 Reputation points

2021-06-29T11:04:03.563+00:00

@Pramod Valavala Does this parameter control the execution of HTTP functions from the request trigger to the returning HTTP response? For instance, if there are some additional background calculations after each HTTP request, will the "maxConcurrentRequests" control the fact end of these calculations, or will it allow running the next HTTP request?
Pramod Valavala 20,656 Reputation points Microsoft Employee Moderator

2021-06-29T12:51:59.703+00:00

@Elizaveta Omelyusik These settings will restrict the HTTP functions alone. While other code apart from your functions will continue to run, this is not recommended as the instance could just be removed without warning when no functions are running.

Instead, you should schedule background jobs via queues and have separate functions to process them.
Elizaveta Omelyusik 26 Reputation points

2021-06-29T14:28:42.143+00:00

@Pramod Valavala Will those queue-triggered functions be executed one by one? Or is there is a way to set a maximum of concurrent executions for that function? There is a "maxConcurrentCalls" parameter but it restricts all functions' execution in a function app.
Pramod Valavala 20,656 Reputation points Microsoft Employee Moderator

2021-07-01T04:45:51.787+00:00

@Elizaveta Omelyusik You can leverage Service Bus Message Sessions to ensure serial execution and ensure to set the isSessionsEnabled configuration on the trigger.

Share via

Azure Function is not creating the new resources and environment for each request when using the docker container

0 additional answers

Your answer