azure/azure-functions-docker

This repo contains the base Docker images for working with azure functions

Build Status:

CoreTools Healthcheck:

Dockerhub

V4 Images

See release artifacts for Dockerfiles - example : https://github.com/Azure/azure-functions-docker/releases/download/3.4.2/3.4.2-appservice.zip

Information about how we keep our images up to date can be found in the wiki

Dotnet

mcr.microsoft.com/azure-functions/dotnet

Linux amd64 Tags

Tags	OS Version
4	Debian 11
4-slim	Debian 11
4-appservice	Debian 11
4-appservice-quickstart	Debian 11

mcr.microsoft.com/azure-functions/dotnet-isolated

Tags	OS Version
4-dotnet-isolated6.0	Debian 11
4-dotnet-isolated6.0-slim	Debian 11
4-dotnet-isolated6.0-appservice	Debian 11
4-dotnet-isolated6.0-appservice-quickstart	Debian 11
4-dotnet-isolated7.0	Debian 11
4-dotnet-isolated7.0-slim	Debian 11
4-dotnet-isolated7.0-appservice	Debian 11
4-dotnet-isolated7.0-appservice-quickstart	Debian 11
4-dotnet-isolated8.0	Debian 12
4-dotnet-isolated8.0-slim	Debian 12
4-dotnet-isolated8.0-appservice	Debian 12

Node

mcr.microsoft.com/azure-functions/node

Linux amd64 Tags

Tags	OS Version
4-node14	Debian 11
4-node14-slim	Debian 11
4-node14-appservice	Debian 11
4-node14-core-tools	Debian 11
4-node14-appservice-quickstart	Debian 11
4-node16	Debian 11
4-node16-slim	Debian 11
4-node16-appservice	Debian 11
4-node16-appservice-quickstart	Debian 11
4-node18	Debian 11
4-node18-slim	Debian 11
4-node18-appservice	Debian 11
4-node18-appservice-quickstart	Debian 11
4-node20	Debian 11
4-node20-slim	Debian 11
4-node20-appservice	Debian 11

Powershell

mcr.microsoft.com/azure-functions/powershell

Linux amd64 Tags

Tags	OS Version
4-powershell7.2	Debian 11
4-powershell7.2-slim	Debian 11
4-powershell7.2-appservice	Debian 11
4-powershell7.2-appservice-quickstart	Debian 11
4-powershell7.4	Debian 12

Java

mcr.microsoft.com/azure-functions/java

Linux amd64 Tags

Tags	OS Version
4-java8	Debian 11
4-java8-slim	Debian 11
4-java8-appservice	Debian 11
4-java8-build	Debian 11
4-java11	Debian 11
4-java11-slim	Debian 11
4-java11-appservice	Debian 11
4-java11-build	Debian 11
4-java17	Debian 11
4-java17-slim	Debian 11
4-java17-appservice	Debian 11
4-java17-build	Debian 11
4-java17-appservice-quickstart	Debian 11
4-java21	Debian 12
4-java21-slim	Debian 12
4-java21-appservice	Debian 12
4-java21-build	Debian 12
4-java21-appservice-quickstart	Debian 12

Python

mcr.microsoft.com/azure-functions/python

Linux amd64 Tags

Tags	OS Version
4-python3.7	Debian 11
4-python3.7-slim	Debian 11
4-python3.7-appservice	Debian 11
4-python3.7-buildenv	Debian 11
4-python3.7-appservice-quickstart	Debian 11
4-python3.8	Debian 11
4-python3.8-slim	Debian 11
4-python3.8-appservice	Debian 11
4-python3.8-appservice-quickstart	Debian 11
4-python3.8-buildenv	Debian 11
4-python3.9	Debian 11
4-python3.9-slim	Debian 11
4-python3.9-appservice	Debian 11
4-python3.9-appservice-quickstart	Debian 11
4-python3.9-buildenv	Debian 11
4-python3.10	Debian 11
4-python3.10-slim	Debian 11
4-python3.10-appservice	Debian 11
4-python3.10-buildenv	Debian 11
4-python3.10-appservice-quickstart	Debian 11
4-python3.11	Debian 11
4-python3.11-slim	Debian 11
4-python3.11-appservice	Debian 11
4-python3.11-buildenv	Debian 11
4-python3.11-appservice-quickstart	Debian 11

Base

mcr.microsoft.com/azure-functions/base

Linux amd64 Tags

Tags	OS Version
4	Debian 11
4-slim	Debian 11
4-appservice	Debian 11

MCR Docs

For new images update the following:

• MCR Syndication

• Docs on DockerHub

Contributing

This project welcomes contributions and suggestions. Most contributions require you to agree to a Contributor License Agreement (CLA) declaring that you have the right to, and actually do, grant us the rights to use your contribution. For details, visit https://cla.microsoft.com.

When you submit a pull request, a CLA-bot will automatically determine whether you need to provide a CLA and decorate the PR appropriately (e.g., label, comment). Simply follow the instructions provided by the bot. You will only need to do this once across all repos using our CLA.

This project has adopted the Microsoft Open Source Code of Conduct. For more information see the Code of Conduct FAQ or contact [email protected] with any additional questions or comments.

azure-functions-docker's Issues

Change default Python image to 3.7

For the Seabreeze platform.

Cannot pass logging envronment variable down to the WebHost

I have tried both ENV settings (It appears that WebHost has been updated here to use the second one):

ENV host:logger:consoleLoggingMode=always
ENV consoleLoggingMode=always

Undoing the following commit seems to allow it to work again:
Commit 36b4bba6457501

Enable remote debugging for Python Functions

Only applicable to Function Apps on dedicated/app service plan and premium plan.

Requirements:

• Container should start in debug mode using $APPSVC_TUNNEL_PORT which is set by app service
• Include pip install --pre ptvsd as part of the dockerfile

More details TBD.

Pre-install popular apt-get packages to the Python image

libpq-dev
python3-dev 
libevent-dev 
unixodbc-dev

@ahmelsayed

python.Dockerfile is using an old version of the worker

It seems python.Dockerfile is pulling in an old version (1.0.0a4) of the worker's source code, is that intentional? If not, please update to use the latest release 1.0.0a6

azure-functions-docker/host/2.0/stretch/amd64/python.Dockerfile

Lines 3 to 13 in 701bb7c

	ARG WORKER_TAG=1.0.0a4
	FROM ${BASE_IMAGE}:${BASE_IMAGE_TAG}
	ARG WORKER_TAG
	RUN apt-get update && \
	apt-get install -y git wget
	RUN wget https://github.com/Azure/azure-functions-python-worker/archive/${WORKER_TAG}.tar.gz && \
	tar xvzf ${WORKER_TAG}.tar.gz && \
	mv azure-functions-python-worker-* azure-functions-python-worker

Install Microsoft odbc driver for sql server in python image

Python function apps using pyodbc to connect to Azure SQL are failing with
('01000', "[01000] [unixODBC][Driver Manager]Can't open lib 'SQL Server' : file not found (0) (SQLDriverConnect)")

Instructions to install:
https://docs.microsoft.com/en-us/sql/connect/odbc/linux-mac/installing-the-microsoft-odbc-driver-for-sql-server?view=sql-server-2017

Portal doesn't display the runtime version

From @davidebbo on April 17, 2018 1:43

It just shows the ~ version but not the actual host version:

The fact that it incorrectly says to upgrade is tracked separately by #2564.

Copied from original issue: Azure/azure-functions-ux#2567

Publish an updated image of microsoft/azure-functions-runtime

Am I in the right repo for this image (microsoft/azure-functions-runtime:2.0.0-nanoserver-1709)?

The latest update was six months ago and because of this issue the host runtime needs to be updated in order to work with Microsoft.NET.Sdk.Functions > 1.0.6.

Create a core-tools tag image

No Console Logging

In the latest build all console logging of errors is gone, and my queueTrigger do not start .
Using microsoft/azure-functions-node8:2.0

'amd64/powershell.Dockerfile' was changed to derive from the 2.0-arm32v7 version of base image

The powershell.Dockerfile under 'amd64' folder was changed by @ahmelsayed to use the 2.0-arm32v7 version of base image while the powershell.Dockerfile under 'arm32v7' also uses 2.0-arm32v7. This doesn't look right to me. Is this change intentional?

See 015b97f#diff-11f1eb5263439c21200d6c06fdf6bc28R2

Azure SDKs native in Functions

Specific customer feedback indicates the experience in Functions can be improved by making Azure SDKs native in Functions (currently requires packaging SDK and uploading). This would create an experience for Functions more aligned to a principle of “inter-connected tissue” to facilitate serverless to hook among the services within Azure.

Docker image size comparison

From @priyaananthasankar on December 14, 2018 8:19

I wrote a forecasting app and launched it in ACI using Python Functions on Container, through func core tools, this uses Blob binding and is triggered through HTTP

func --init Docker and compiled the docker image. The code for this is here:

https://github.com/priyaananthasankar/parseltongue/tree/master/PyFunctionsOnACI/ForecastingApp

I also wrote the exact same forecasting app in my own custom container which uses Azure Blob Storage for Python + Flask for HTTP Trigger (basically does not use the functions programming model)

I see the image using functions programming model is almost 400MB more than the custom image. This is a case for running functions on AKS or ACI etc, the larger the container the more time it takes to spin it up. There must be a better case for containerized app that uses the Python functions programming model.

Check out the comparison here: https://github.com/priyaananthasankar/parseltongue/tree/master/Comparisons

Copied from original issue: Azure/azure-functions-python-worker#265

Support Python 3.7

We need a docker image to support Functions written using Python 3.7.

Install the Python language worker and related dependencies in the base image

In Azure, we want to use a version of the language worker installed in the base image that we can update and maintain on behalf of the user.

/cc @ahmelsayed

Publish PowerShell image to MCR

Hook up the build pipeline for the new PowerShell image and publish it to the MCR repo. @ahmelsayed

Update ARM32 Linux and Windows containers with breaking changes

Since the big breaking change in V2, these two images haven't been updated. This is blocking a partner team.

Functions v1 image

I've seen references to Functions v1 being run inside a container,

In this version we added support for:
Functions v1 (.NET Framework) running in Windows Server Core container
Azure/Azure-Functions#288 (comment)

is there any chance that Dockerfile can be added to that repo and/or that image could be published so we can use it as part of our test environment for our v1 Functions till v2 is out of preview?

Remove dependency on worker.py

Currently, there is a dependency on worker.py being present in the Python worker. However, this would not necessarily be the case in the future e.g. running the language worker module directly. Running the worker process should rely on configuration rather than the presence of a particular file.

python image is too large

Using with Docker Compose

Here's my docker-compose.yml;

version: '3'

services:
  func:
    environment:
      AzureWebJobsScriptRoot: /www/src
      AzureFunctionsJobHost__Logging__Console__IsEnabled: 'true'
    image: mcr.microsoft.com/azure-functions/node
    ports:
      - 7071:80
    restart: always
    volumes:
      - .:/www
    working_dir: /www

  mongo:
    image: mongo:3.4
    restart: always

Four questions;

1. Is the behavior around AzureFunctionsJobHost__Logging__Console__IsEnabled sorted yet? Is there a cleaner flag I can set to enable verbose'ish logging?
2. When I docker-compose up, my app cannot access the environment variables in local.settings.json. How can I get it to load variables from there?
3. This app always boots in production mode; Hosting environment: Production. How can I get to run in development mode so NODE_ENV is set to development and there's more logging, etc? Couldn't find any documentation on this. Hopefully, this will also enable the use of console.log as right now only context.log is available.
4. How can I get it to work with authLevel set to function, which is how I want my code to be deployed to Azure? Will setting the environment to development fix that? Or will I have to commit to source control with anonymous and as part of my build process, change to function?

cc @mhoeger @paulbatum

Install pyodbc in the base Python image

Post pip install azure-common, worker module imports are failing

Base image: microsoft/azure-functions-pyon3.6:v2.0.11737-alpha

Repro:

After pip installing the azure-common module, I get the following error when trying to import azure.functions. ModuleNotFoundError: No module named 'azure.functions'

$ docker run -it microsoft/azure-functions-pyon3.6:v2.0.11737-alpha bash
root@17dd101117b3:/# pip install azure-common
Collecting azure-common
  Downloading https://files.pythonhosted.org/packages/97/3b/2c7cda25382c3bb566008c5c8f8aa28663fd15a80a6204c76ae0035de107/azure_common-1.1.11-py2.py3-none-any.whl
Collecting azure-nspkg>=2.0.0 (from azure-common)
  Downloading https://files.pythonhosted.org/packages/cd/a0/76ca6659ae9afd7567fdbb5b9c85e9c9a0b48238cfcacd92525484408f66/azure_nspkg-2.0.0-py2.py3-none-any.whl
Installing collected packages: azure-nspkg, azure-common
Successfully installed azure-common-1.1.11 azure-nspkg-2.0.0
You are using pip version 9.0.1, however version 10.0.1 is available.
You should consider upgrading via the 'pip install --upgrade pip' command.
root@17dd101117b3:/# python
Python 3.6.4 (default, May  1 2018, 23:33:54)
[GCC 6.3.0 20170516] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import azure.functions
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ModuleNotFoundError: No module named 'azure.functions'
>>> 

Workaround to get azure.functions import to work

Add the following lines to the docker file. This seems to change how the azure.functions module is installed. Consequently, I can import the azure.functions again.

pip uninstall azure-functions-python
pip install git+https://github.com/Azure/[email protected]#egg=azure-functions-python

Other errors

However, additional errors are logged by worker in the console.

Docker file

FROM microsoft/azure-functions-python3.6:v2.0.11737-alpha

ENV host:logger:consoleLoggingMode=always

COPY . /home/site/wwwroot

RUN pip uninstall azure-functions-python -y

RUN pip install git+https://github.com/Azure/[email protected]#egg=azure-functions-python

RUN pip install azure-common

info: Worker.python.e27f3d6b-6ca2-4939-9cf9-df86628d4007[0]
      ImportError: cannot import name 'bindings'
fail: Worker.python.e27f3d6b-6ca2-4939-9cf9-df86628d4007[0]
      Worker encountered an error.

...

fail: Function.TimerTrigger[0]
      => System.Collections.Generic.Dictionary`2[System.String,System.Object]
      System.Private.CoreLib: Exception while executing function: Functions.TimerTrigger. System.Private.CoreLib: One or more errors occurred. (Worker process with pid 39 exited with code 1) (Worker process with pid 89 exited with code 1) (Worker process with pid 135 exited with code 1). Worker process with pid 39 exited with code 1.
fail: Function.TimerTrigger[0]
      => System.Collections.Generic.Dictionary`2[System.String,System.Object]
      Executed 'Functions.TimerTrigger' (Failed, Id=f20f345a-56da-4415-9e14-71e0ca6fb5c1)
Microsoft.Azure.WebJobs.Host.FunctionInvocationException: Exception while executing function: Functions.TimerTrigger ---> System.AggregateException: One or more errors occurred. (Worker process with pid 39 exited with code 1) (Worker process with pid 89 exited with code 1) (Worker process with pid 135 exited with code 1) ---> System.Exception: Worker process with pid 39 exited with code 1
   --- End of inner exception stack trace ---
   at System.Runtime.ExceptionServices.ExceptionDispatchInfo.Throw()
   at System.Runtime.CompilerServices.TaskAwaiter.HandleNonSuccessAndDebuggerNotification(Task task)
   at System.Runtime.CompilerServices.TaskAwaiter`1.GetResult()
   at Microsoft.Azure.WebJobs.Script.Description.WorkerLanguageInvoker.<InvokeCore>d__6.MoveNext() in /azure-functions-host-2.0.11737-alpha/src/WebJobs.Script/Description/Rpc/WorkerLanguageInvoker.cs:line 73
--- End of stack trace from previous location where exception was thrown ---

Create ssh enabled images for use in AppService

The changes needed for .Net are
c6beee8

(For node we might need to launch the node process in debug mode additionally)

We will need to have separate ssh enabled images ( dotnet , node , python to begin with) to avoid security issue with having public username/password in the dockerfile. When used in appservice this is ok since the port is only exposed through kudu which requires auth.

The images will be named with -xdebug as part of their name (Ex: appsvc/php:5.6.38-apache-xdebug_1810291954). When remote debugging is enabled on a site , LWAS will pick the -xdebug image instead of regular image.

Matching launch.json config to use with vs code

{
"name": ".NET Core SSH Attach",
"type": "coreclr",
"request": "attach",
"processId": "${command:pickRemoteProcess}",
"pipeTransport": {
"pipeProgram": "ssh",
"pipeArgs": [ "-T", "[email protected]", "-p", "PORT_NUMBER", "-i", "PATH_TO_SSHPRIVATEKEY" ],
"debuggerPath": "/root/vsdbg/vsdbg",
"pipeCwd": "${workspaceRoot}",
"quoteArgs": true
}
}
debugger path is preset based on dockerfile debugger installation location.
PORT_NUMBER is from
az webapp remote-connection create -g RESOURCE_GROUP -n APP_NAME -p PORT_NUMBER

SSH Connection fails in function apps

We might have to install the ssh on function app container.

The links below might be helpful,

1. https://github.com/Azure-App-Service/node/blob/master/10.1.0/Dockerfile
2. https://github.com/Azure-App-Service/node/blob/master/10.1.0/sshd_config

Provide Node.js 10 worker runtime

The docs on docker hub say that azure-functions/node:2.0 has node 10 worker runtime

but it is node 8 actually according to the docker file https://github.com/Azure/azure-functions-docker/blob/master/host/2.0/stretch/amd64/node.Dockerfile#L8

Deployment Steps/Script for Linux/Seabreeze

As we move closer to Python GA in June, there is a critical need to test Python scenarios directly on the Seabreeze Linux platform using our Docker container images. There are crucial development cases that require the use of Seabreeze specific testing:

1. Logging to Kusto/MDS/AI
2. Cold start perf testing/ profiling
3. Integration testing

The ask here is to have a set of steps for the following workflow: (1) Host + Worker Code --> (2) Deploy to Seabreeze in Docker Container --> (3) Check Logs/ Peform perf testing/ Use container for integration testing

Ideally, (1) and (2) above could be automated as part of a script so that we are able to point to a worker and host repo and then deploy the full package for testing in Seabreeze.

Please feel free to share existing scripts that we can use for this. Thanks!

Where is the new nanoserver variant?

I was told that there is a new location for the images:

@NickDarvey That image is being replaced with these ones (pick your language runtime)
azure-functions-dotnet-core2.0
azure-functions-node8
azure-functions-python3.6

Azure/azure-functions-core-tools#335 (comment)

However, I can't find a nanoserver variant which was available as part of the old image. Where can I find it?

Update pip version for the Python base image

Azure portal cannot list my functions anymore

Hello,

Azure portal is not listing my functions anymore. May it comes from recent changes ?

When azure portal tries to list my functions in the interface, it tries to access /admin/functions and here is the result :

OPTIONS https://xxxxxxx.azurewebsites.net/admin/functions 404 ()
Failed to load https://xxxxxxx.azurewebsites.net/admin/functions: Response to preflight request doesn't pass access control check: No 'Access-Control-Allow-Origin' header is present on the requested resource. Origin 'https://functions.azure.com' is therefore not allowed access.
POST https://functions.azure.com/api/passthrough 500 (Internal Server Error)

I am not sure if it may comes from the docker image.
I am using 2.0 tag. Does the image behind this tag change overtime ? If yes it would be great to have immutable docker image tag.

Docker image is from microsoft/azure-functions-base:2.0

Functions With Storage Queue Triggers

I have used this base image to run a function which uses a storage queue trigger.
This works when I run the container on my machine:
docker run --name function functionimagename
as in I can see then when putting messages on the relevant queue, they are consumed.

However, if I deploy to Azure (as a Functions app with "OS"=="Docker"), I can see that a container is run using my image, but items on the queue are not consumed.

Seeing as the issue here is that the function is not being triggered, I am struggling to diagnose why it is not being triggered.

I have made a sample project reproducing this behaviour
https://github.com/thyde1/docker-azure-functions-test

Any ideas why this isn't working in Azure would be greatly appreciated.

Support .NET Core 2.2 SDK as Base Image

Is there a timeline posted for migrating to 2.2-sdk as the base image(s)?

We would love to upgrade to .NET Core 2.2 across our platform, but we need parity across all of our images for library/package compatibility, etc.

Also, how does the azure-functions-dotnet-core2.0 naming convention fit in with the .NET Core SDK naming convention? Is 2.0 here referencing Azure Functions 2.0 or .NET Core SDK? It's somewhat confusing based on the fact that the host is based on the 2.1 SDK image currently 😜

Change image layer order to minimize changes on each update

Microsoft ODBC Driver 17 for SQL Server - Python

I'd like to be able to connect from a Python Azure Function in the 2.0 runtime (consumption plan) to an Azure SQL Server 17 database.

Currently receiving the error: OdbcNoLibrary: ODBC Library is not found. which should be resolved by #32, but I believe the correct driver would also need to be installed for unixodbc to actually make the connection. I found instructions for that here: https://docs.microsoft.com/en-us/sql/connect/odbc/linux-mac/installing-the-microsoft-odbc-driver-for-sql-server?view=sql-server-2017 .

Could we have this driver installed to the base image for Python Azure Functions, to enable out-of-the-box SQL Server 17 connectivity?

Create separate images for Python, Node and C#

We need separate images containing the functions host, language runtime and the language worker.

Install Azure SDK into base python image

It would be nice to add pip install azure to https://github.com/Azure/azure-functions-docker/blob/master/host/2.0/stretch/amd64/python3.6/Dockerfile

Of course it further aggravates #2 (although not sure if important)

This is comparable to what AWS does in Lambda, where 100% of the AWS SDK is available.

It also reduces lots of pain on the developer. Packaging a function is usually not too big of a deal, but the Azure SDK is a very complex package (dependency tree) with lots of natively compiled binaries in it, it is nearly impossible to package without the complex docker based process used in the Functions CLI - and even then you end up with a very large zip and a slow packaging process.

I'm happy to make a PR and test the image/size if there is interest.

HttpTrigger Local Function Returns 406 in the Browser

From @aluong on May 8, 2018 22:13

I cloned and the example which targets base image: microsoft/azure-functions-python3.6:v2.0.11737-alpha

When I go to

http://localhost:8080/api/HttpTrigger?name=hi

I see an Http Error 406. In the docker logs I see that the HttpTrigger was hit.

info: Function.HttpTrigger[0]
      => System.Collections.Generic.Dictionary`2[System.String,System.Object] => System.Collections.Generic.Dictionary`2[System.String,System.Object]
      Executing 'Functions.HttpTrigger' (Reason='This function was programmatically called via the host APIs.', Id=f7d69d25-233a-41da-a569-95a2bdaa9ca1)

Copied from original issue: Azure/azure-functions-docker-python-sample#15

[Python] Install required packages to the virtual environment

Context: Creating the deployment package (BYOF)

We need to install the user's packages and dependencies to the virtual environment in production. For this, we will pip install using the requirements.txt manifest and pre-built archives in the wheels folder.

pip install --no-index --find-links=/functionapp/wheels -r requirements.txt

The proposal is to do this as part of the start.sh script before activating the language worker.

Remove extra platforms runtimes from powershell worker.

see Azure/azure-functions-core-tools#1141 for details

Add python alpine image to public container registry

This is probably related to #2. I have seen that we have a alpine Dockerfile for python, but this image is not being built and pushed to the registry.

Can this be added to the build pipeline, please? I have already been using since functions v2.0.12285 without any problem.

install msodbcsql in the python image

func azure functionapp publish --build-native-deps fails with docker error

Hi All!
I am tryting to publish a python function but I got an error message.
I used this command:
func azure functionapp publish sanyifunction --build-native-deps

Environment is Centos7 and Python 3.6.7

This is the error message:

Getting site publishing info...
Running 'docker pull mcr.microsoft.com/azure-functions/python:2.0.12285'..done
Running 'docker run --rm -d mcr.microsoft.com/azure-functions/python:2.0.12285'..done
Running 'docker exec -t 088384 mkdir -p /home/site/wwwroot/'..done
Running 'docker cp /tmp/15pjtmni.l2t/. 088384:/home/site/wwwroot'..done
Running 'docker kill 088384'..done

Error running docker cp /tmp/15pjtmni.l2t/. 08838429b9ba5e734c977c03f3008ff86b6616f7f66c68063d3bafd3fc63a6dc:/home/site/wwwroot.
output:
Error: No such container:path: 08838429b9ba5e734c977c03f3008ff86b6616f7f66c68063d3bafd3fc63a6dc:/home/site

This is the requirements.txt

cat requirements.txt
azure-functions==1.0.0b3
azure-functions-worker==1.0.0b3
grpcio==1.14.2
grpcio-tools==1.14.2
protobuf==3.6.1
six==1.12.0
azure
azure-mgmt-resource
azure-datalake-store
azure-mgmt-datalake-store
azure-mgmt-datalake-analytics

The code is running locally. Also I have to use the --build-native-deps otherwise I get this error message:

There was an error restoring dependencies.ERROR: cannot install cryptography-2.5 dependency: binary dependencies without wheels are not supported. Use the --build-native-deps option to automatically build and configure the dependencies using a Docker container. More information at https://aka.ms/func-python-publish

Can it be a bug? Or maybe I am doing something wrong?

Help me Understand the Functions Runtime and Pooling.

I did a pull of the azure functions base. the image size is a whopping 3.92 gb.
mcr.microsoft.com/azure-functions/base latest c1f034e6ab2e 1 hours ago 3.92GB

I thought this might be the runtime, which means this is the image that actually manages the pooling of Functions, and orchestrates, shutting down, serving etc.

However it appears this image serves as the base for all individual functions.

Can someone please help me understand more on this topic? Is it true that every function that gets spun up to serve requests is actually at least 3.92 gb!!!?!?!

If I am incorrect on this could I get some detail as function lifetime flow and how the azure runtime will orchestrate these functions?

We are looking into azure functions, and if this runtime is open source, and supported we could always port it to work on kubernetes if there were ever a need to switch cloud vendors. The way I currently understand functions and how they work, and then reviewing these images, is a step back in the direction for azure functions.

Support Anaconda in-built image

• Installs some of the best and most used libraries by default
• Easy to install modules with conda compared to pip(+wheels)
• Most of engineers running python on functions v2 might use it for data processing with pandas or other libraries. they might not be interested in going through extra steps for installing them..

Minor update to Python image

As per official Python docker image, could we add some important environment variables:

ENV LANG C.UTF-8
ENV PYTHON_VERSION 3.6.6
ENV PYTHON_PIP_VERSION 18.0

Also, while at it, why don't we update to version 3.6.6. I have been doing some tests and, so far, not a single problem.

documentation

Please provide more information and/or documentation for reference deployments and how to use the docker images for no premises deployments/development

Update python3.6 dockerfile to pull python worker from a specific tag

As we are in private preview, need to update docker file to pull Python Worker from a tagged build

Python requirement absl-py installation fails (binary dependencies without wheels are not supported)

When trying to deploy a function app that requires TensorFlow (e.g., tensorflow==1.11.0 in requirements.txt), the following error occurs:

$ az functionapp createpreviewapp -n Fun2xApp -g XXXXXXX -l "WestUS" -s  XXXXXXX --runtime  python --is-linux
$ func azure functionapp publish Fun2xApp  --force
Getting site publishing info...
pip download -r /Users/patryk.laurent/Desktop/TEST/azure_functions2x/MyFunctionProj/requirements.txt --dest /var/folders/ny/48h0ry5j5bdcvv99jqxs2b24qrj203/T/azureworker13qtj_47
pip download --no-deps --only-binary :all: --platform manylinux1_x86_64 --python-version 36 --implementation cp --abi cp36m --dest /var/folders/ny/48h0ry5j5bdcvv99jqxs2b24qrj203/T/azureworkerf2kx2ew_ astor==0.7.1
pip download --no-deps --only-binary :all: --platform manylinux1_x86_64 --python-version 36 --implementation cp --abi cp36m --dest /var/folders/ny/48h0ry5j5bdcvv99jqxs2b24qrj203/T/azureworkerf2kx2ew_ azure_functions_worker==1.0.0a4
pip download --no-deps --only-binary :all: --platform manylinux1_x86_64 --python-version 36 --implementation cp --abi cp36m --dest /var/folders/ny/48h0ry5j5bdcvv99jqxs2b24qrj203/T/azureworkerf2kx2ew_ setuptools==40.5.0
pip download --no-deps --only-binary :all: --platform manylinux1_x86_64 --python-version 36 --implementation cp --abi cp36m --dest /var/folders/ny/48h0ry5j5bdcvv99jqxs2b24qrj203/T/azureworkerf2kx2ew_ azure_functions==1.0.0a4
pip download --no-deps --only-binary :all: --platform manylinux1_x86_64 --python-version 36 --implementation cp --abi cp36m --dest /var/folders/ny/48h0ry5j5bdcvv99jqxs2b24qrj203/T/azureworkerf2kx2ew_ Werkzeug==0.14.1
pip download --no-deps --only-binary :all: --platform manylinux1_x86_64 --python-version 36 --implementation cp --abi cp36m --dest /var/folders/ny/48h0ry5j5bdcvv99jqxs2b24qrj203/T/azureworkerf2kx2ew_ Markdown==3.0.1
pip download --no-deps --only-binary :all: --platform manylinux1_x86_64 --python-version 36 --implementation cp --abi cp36m --dest /var/folders/ny/48h0ry5j5bdcvv99jqxs2b24qrj203/T/azureworkerf2kx2ew_ protobuf==3.6.1
pip download --no-deps --only-binary :all: --platform manylinux1_x86_64 --python-version 36 --implementation cp --abi cp36m --dest /var/folders/ny/48h0ry5j5bdcvv99jqxs2b24qrj203/T/azureworkerf2kx2ew_ grpcio_tools==1.14.2
pip download --no-deps --only-binary :all: --platform manylinux1_x86_64 --python-version 36 --implementation cp --abi cp36m --dest /var/folders/ny/48h0ry5j5bdcvv99jqxs2b24qrj203/T/azureworkerf2kx2ew_ tensorboard==1.10.0
pip download --no-deps --only-binary :all: --platform manylinux1_x86_64 --python-version 36 --implementation cp --abi cp36m --dest /var/folders/ny/48h0ry5j5bdcvv99jqxs2b24qrj203/T/azureworkerf2kx2ew_ tensorflow==1.10.0
pip download --no-deps --only-binary :all: --platform manylinux1_x86_64 --python-version 36 --implementation cp --abi cp36m --dest /var/folders/ny/48h0ry5j5bdcvv99jqxs2b24qrj203/T/azureworkerf2kx2ew_ termcolor==1.1.0
pip wheel --no-deps --no-binary :all: --wheel-dir /var/folders/ny/48h0ry5j5bdcvv99jqxs2b24qrj203/T/azureworker8feia_e1 termcolor==1.1.0
pip download --no-deps --only-binary :all: --platform manylinux1_x86_64 --python-version 36 --implementation cp --abi cp36m --dest /var/folders/ny/48h0ry5j5bdcvv99jqxs2b24qrj203/T/azureworkerf2kx2ew_ numpy==1.14.5
pip download --no-deps --only-binary :all: --platform manylinux1_x86_64 --python-version 36 --implementation cp --abi cp36m --dest /var/folders/ny/48h0ry5j5bdcvv99jqxs2b24qrj203/T/azureworkerf2kx2ew_ grpcio==1.14.2
pip download --no-deps --only-binary :all: --platform manylinux1_x86_64 --python-version 36 --implementation cp --abi cp36m --dest /var/folders/ny/48h0ry5j5bdcvv99jqxs2b24qrj203/T/azureworkerf2kx2ew_ six==1.11.0
pip download --no-deps --only-binary :all: --platform manylinux1_x86_64 --python-version 36 --implementation cp --abi cp36m --dest /var/folders/ny/48h0ry5j5bdcvv99jqxs2b24qrj203/T/azureworkerf2kx2ew_ wheel==0.32.2
pip download --no-deps --only-binary :all: --platform manylinux1_x86_64 --python-version 36 --implementation cp --abi cp36m --dest /var/folders/ny/48h0ry5j5bdcvv99jqxs2b24qrj203/T/azureworkerf2kx2ew_ absl-py==0.6.1
pip wheel --no-deps --no-binary :all: --wheel-dir /var/folders/ny/48h0ry5j5bdcvv99jqxs2b24qrj203/T/azureworkerzkuf3tru absl-py==0.6.1

There was an error restoring dependencies.ERROR: cannot install absl-py-0.6.1 dependency: binary dependencies without wheels are not supported.  Use the --build-native-deps option to try building the binary dependencies using a Docker container.

Is there a workaround?

Thank you.

Update runtime to 2.0.12332

Runtime 2.0.12332 is rolling out: Azure/app-service-announcements#159
This is tracking item to update docker images.

Functions 2.x - How to register bindings?

For functions 2.x, it is requirement to explicitly register any extensions aside from HTTP and Timer. I've found that when running a simple ServiceBusTrigger'd function app within a Docker container, the host fails to load the functions runtime because the required extension isn't installed. It seems to not be enough to have the particular extension NuGet referenced in the function source; it has to be installed into the runtime.

I'm thinking I should install the functions core tools in the image and then run "func install extensions" at the entrypoint. Am I missing a more obvious solution?