kubeless can not download function from url which contains &

[yvz5]

Is this a BUG REPORT or FEATURE REQUEST?: BUG

What happened: I have created an azure blob storage and uploaded the function as a zip file. created the sas token and got a link which includes & signs. e.g. azure.blob.com/function.zip?hi&hello=world

What you expected to happen: that kubeless escapes the curl command inside the prepare container

How to reproduce it (as minimally and precisely as possible): just use any URL with & in it

Anything else we need to know?:

Environment:

• Kubernetes version (use kubectl version):
• Kubeless version (use kubeless version):
• Cloud provider or physical cluster: azure

So this is really weird because I would expect the curl command to be escaped.

If I url encode the link and replace & with %26 this time I get resource is not found from azure and checksums wont match.

the arguments of the prepare container:

 Args:
      curl https://xxx.blob.core.windows.net/functions/function.zip?sp=r&st=2019-10-30T14:29:28Z&se=2019-12-01T22:29:28Z&spr=https&sv=2019-02-02&sr=b&sig=xxx%3D -L --silent --output /tmp/func.fromurl && echo '71aecbb768cc278a590c4dbc667531b385bf88eb56ab3dc77558dd52eace4cd8  /tmp/func.fromurl' > /tmp/func.sha256 && sha256sum -c /tmp/func.sha256 && unzip -o /tmp/func.fromurl -d /kubeless && cp /src/package.json /kubeless

logs from prepare container:

C:\src\deployment>kubectl logs fn-6b87fd859-fm28p -c prepare
sh: -L: not found

[andresmgot]

I see the issue, yes we are not escaping those characters.

One question, if you leave the URL with & and you edit the deployment to quote (with simple quotes
like 'https://xxx.blob.core.windows.net/functions/function.zip?sp...'), does it work? That should prevent the container from interpreting the &. If that is like that we can submit a simple patch.

[yvz5]

I am not able to test it as I go to vacation tomorrow. But it would probably solve the problem.

[yvz5]

I did put the ' sign in the fucntion definition and it did work. could you make the changes ?

[andresmgot]

The changes are already in master, if you want to run the latest image before there is a new release you can do so executing:

kubectl set image -n kubeless deployment/kubeless-controller-manager kubeless-function-controller=docker.io/kubeless/function-controller:build-7d305815-8d7f-43e8-a0ac-bbc4899f1761