I'm looking at using ArcGIS Python API in an AWS Lambda function. I have installed all of the minimum dependencies based on Install and Setup | ArcGIS API for Python but some of these dependencies seem to be an overkill when used in AWS. Would the kerberos & ntlm packages actually be required and probably the same with oauthlib?
At the moment, I leaning towards ujson, six & requests are probably the bare minimum for AWS Lambda.
It would be great to find out the actual bare minimum per scenario (eg. If server/portal using local win accounts then only ujson, six & requests are required, etc)
Thanks for any help with this.
I've run into the Lambda deployment size limitation in the past, so I understand your predicament. What functionality do you need from the Python API? On at least one occasion, it was easier to just write a custom class for my Lambdas that handled the essentials. I was doing simple stuff like querying a service to generate a report.
The use case I'm looking at this for CI/CD pipeline. Storing artifacts in S3 bucket (eg. sd files, mapx files, json for web maps) and then invoking lambda via a schedule to take these artifacts from S3 and publish them into ArcGIS Enterprise.
I'm thinking of building a docker file and using it that way instead now.
At the moment, I'm getting an error in loading the packages 'asn1' cannot be found for the crypto package