Helpful guides and examples to make life easier.
The following is a brief series of steps that should help get people up and running with AWS.
As useful as AWS is, the developer community considers it a necessary evil. Rightfully so, as using it can be painstaking and full of errors before the first sign of success. Major aspects of using AWS rely on setup and configuration: creating a root account, IAM users and groups, security credentials, and so on. Once these are set, and with well-defined naming conventions, the administrative evil can be somewhat bypassed with future use.
Use IAM Users & Groups to provision permissions and policies. Otherwise, root access could come in harm’s way.
Aside from IAM, roles are going to be necessary to attach to services such as AWS Lambda. In particular, these roles will also have attached policies, which will allow them to interact with various parts of the AWS ecosystem.
In order for AWS Lambda to use encrypted environment variables with AWS Key
Management Service (KMS), Lambda functions will require roles with policies
that are VPC-enabled. For instance, AWSLambdaVPCAccessExecutionRole is one
such policy. Otherwise, any function call with decryption will timeout and
throw an error.
Note the use of a named profile for credentials associated directly with a particular project or group (i.e., the lab). This configuration is made via
aws configure --profile profilename
where profilename is replaced with the intended name of that user profile for AWS access configuration.
You can both check for existence and view for certainty the configuration and credentials (do NOT make changes to these files directly):
open ~/.aws/config
open ~/.aws/credentials
Creating your first bucket from the command line, assuming proper permissions,
should be straightforward with aws s3api:
aws s3api create-bucket --bucket mytest-bucket
where mytest-bucket is the name of the bucket you’d like to create; the
bucket name must be unique across the AWS ecosystem, so if you are returned
an error about the bucket name, make it unique. The bucket is
made at the default location of your AWS profile configuration. If you’d like
to specify a location, include LocationConstraint=location as an option on
the command. The location can be one of these
regions under the ‘Code’ column. A succesful response of this command to create buckets will look like
{
"Location": "/mytest-bucket"
}
To upload a local file, run
aws s3 cp <file-name-with-extension> s3://mytest-bucket
The command starts with making sure the aws tool is being used, specifies
that the service is s3, and uses cp to copy the path, in this case a file,
to the S3 bucket “mytest-bucket”. If such a file is too large to directly upload
to AWS, first compress the file with
zip -r9 <file-name>.zip <file-name-with-extension>
and then run
aws s3 cp <file-name>.zip s3://mytest-bucket
where <file-name>.zip is the compressed file. Before running this,
you might include the option --dryrun to be sure it works as intended.
Important: In order to use AWS Lambda with package dependencies, such as
the tweepy package in Python, the command line interface aws must be installed on your
local machine. This is non-negotiable both for security purposes and for
package deployment. If you have not configured your CLI, please scroll back up
and do so before continuing on this section.
Nothing comes easy with AWS, and using Python packages is no exception to this
rule. To install pandas and numpy, an AWS Linux compatible version of each
must be compiled. A great tutorial on doing so with Docker is found here
and references these useful public Docker images.
The SAM CLI is a very convenient method of creating, organizing, building and deploying Lambda functions. Start by installing it.
To initialize a new Lambda function, go to the directory you’d like to create it in and run
sam init -r python3.8 -d pip
where the runtime is specified as Python 3.8 by -r python3.8 and dependency manager as pip -d pip.
This initialization will prompt with some interactive questions, including whether you’d like to use a quick start template from AWS or a custom template at a specified location (i.e., in a GitHub repository).
While a name for the application is requested in the set of prompts, you can preemptively name it by extending the above initialization command with `-n