numeract / aws-lambda-r Goto Github PK

TODO:

• bring in scripts
• minimal R setup

I assume that the R script in the Lambda has access to the AWS OS it's running on so theoretically it could use for example Clodyr's R library aws.s3 and fetch files from s3.

# get file as raw vector
get_object("mtcars.Rdata", bucket = "my_bucket")

Then you don't have to pass it thru Python and also you avoid rewriting your existing R code base. Or am I missing something? Is this not possible?

• Add script in order to allow API Gateway to be accessed by CloudWatch

Problem

• the scripts rely on certain AWS roles, lambda authentication function etc (one time setup)

TODO

• decide what to automate
  • e.g., no need to create VPC and subnet --> write instructions
• one-time scripts to create roles (everything related to IAM)
  • instructions / doc / script how to remove
• one-time scripts to setup lambda
  • instructions / doc / script how to remove
• one-time scripts to setup API gateway
  • instructions / doc / script how to remove

Hello,

Very nice job!

I'm confused about EC2. Reading your documentation, I can't figure out why do I need a EC2 instance to build the package. Is it possible to do that locally? Is there some reason for that?

As I understand, after setting up all variables, I just need to run 01_main.sh script, right?

Best regards,
Clovis

Problem:

• if R fails to install any package, the script does not stop

TODO

• check if installation of R packages was successful; fail if any package fails installation

Problem:

• script continues if yum install fails
• no option in settings to specify what to install

TODO:

• put yum packages in an array in settings
• read array, generate install commands
  • best if we have only one yum install command
• check for errors and fail if error after install

Hi

in script 24-setup_lambda.sh I cant seem to be able to create the Lambda Authorizer Function.
It fails with the following (the permissions for the IAM user running this are full access to lambda)

INFO : Creating Lambda Authorizer Function ...

An error occurred (InvalidParameterValueException) when calling the CreateFunction operation: The role defined for the function cannot be assumed by Lambda.

An error occurred (ResourceNotFoundException) when calling the GetFunctionConfiguration operation: Function not found: arn:aws:lambda:ap-southeast-2:11111111111:function:aws-lambda-r-lambda-authorizer
ERROR: Failed to obtain ARN of Lambda Authorizer Function aws-lambda-r-lambda-authorizer. Exiting.```

Problem

• there is no checking of the file size, it may be too big 250MB to run on Lambda
• the script does not stop

TODO

• check total deployment package size (outside zip)
• fail if too large

Problem

• if we have scripts that create resources on AWS (e.g. AMIs), we need to save their IDs

Solution

• read & write from a user specific .ini file (instead of executing a user specific .sh file)
• parsers:
  • https://github.com/albfan/bash-ini-parser
  • https://github.com/rudimeier/bash_ini_parser (preferred because the licence is specified)

Note: remember to add the third party references in README and NOTICE files, see

• https://github.com/rstudio/rstudio/blob/master/NOTICE
• https://softwareengineering.stackexchange.com/questions/136400/if-i-use-multiple-third-party-components-using-several-different-open-source-li

Problem:

• lambda deployment too slow

Solution:

• use 2 AMI: amazon & project

TODO:

• new feature branch
• in default settings, use 2 variabiles
  • ALX_AMI_ID="ami-657bd20a" : last version of Amazon Linux, w/o updates
  • EC2_AMI_ID=""
• if EC2_AMI_ID not defined or there is no AMI with this ID -->
  • check if there is a file settings/settings_ami.txt and whether it contains a string --> EC2_AMI_ID=string
  • again, if EC2_AMI_ID not defined or there is no AMI with this ID --> EC2_AMI_ID=""
• if EC2_AMI_ID is defined --> done, skip the steps below
• consider options not to update with yum or update R packages
• a script based on scripts 11 and 13:
  • if ALX_AMI_ID definit but no AMI --> exit 1
  • start with ALX_AMI_ID + update + yum install + virtualenv + pip + cp + cp + install R packages + cp
  • use $PRJ_NAME
  • do not copy files from project; do not create zip
  • create an AMI from this instance
  • write AMI ID in file settings/settings_ami.txt
  • terminate this EC2
• main script: check if file exists settings/settings_ami.txt and if it contains a string --> EC2_AMI_ID=string
  • if EC2_AMI_ID not defined or no AMI --> exit 1
• deployment package
  • connect to EC2_AMI_ID
  • copy files (script 06)
  • create zip

Problem

• not everybody wants to use API Gateway; lambda is sufficient

TODO

• split Lambda and API script into two
• option in settings to (or not to) continue with API Gateway setup
• test of lambda function from AWS CLI

On the Amazon AMI I can install from inside R, but then I have this problem:
Installing package into ‘/usr/lib64/R/library’
(as ‘lib’ is unspecified)
Warning in install.packages("jsonlite", repos = "http://cran.r-project.org") :
'lib = "/usr/lib64/R/library"' is not writable
Would you like to use a personal library instead? (y/n)

I can install it this way also:
wget https://cran.r-project.org/src/contrib/jsonlite_1.5.tar.gz
sudo R CMD INSTALL jsonlite_1.5.tar.gz

which probably is the best way, or run R as sudo..

Problem

• no ""serious" example for this solotion

Hi

running through a test deployment after setting up AWS, and am finding this:

INFO : Copy files to local lambda/
INFO : Copy file: python/lambda_get.py to lambda/
INFO : Copy file: python/lambda_post.py to lambda/
INFO : Copy file: example.R to lambda/
INFO : Remove EC2 directory ~/rtest-aws
INFO : Copying local lambda/ to EC2 ~/rtest-aws ...
.gitignore                                                                                                                                                                                                                                                                                                                                100%   84     2.8KB/s   00:00
example.R                                                                                                                                                                                                                                                                                                                                 100% 1281    42.3KB/s   00:00
lambda_get.py                                                                                                                                                                                                                                                                                                                             100%  672    22.2KB/s   00:00
lambda_post.py                                                                                                                                                                                                                                                                                                                            100%  673    22.3KB/s   00:00
README.md                                                                                                                                                                                                                                                                                                                                 100%  267     8.8KB/s   00:00
INFO : Configure EC2, create Lambda package/function & API method ...
INFO : Configure AWS on EC2
INFO : Check AWS configuration on EC2 ...
INFO : AWS Account ID: ********3729
INFO : Lambda Function Name: rtest-aws-master-alpha-resource-v1-get
INFO : Creating deployment package
INFO : PWD: /home/ec2-user/rtest-aws
INFO : Transferring Python 3.6 packages to the deployment package
bash: line 392: /home/ec2-user/env/bin/activate: No such file or directory
ERROR: Cannot create Lambda package/function & API method. Terminating end exiting ...

one thought is that perhaps my custom ami is not right, or perhaps a bug somewhere?

Problem

• we have only one list of packages to install from CRAN

TODO

• another array option with a list of packages to install from github
  • need to use name/repo format
• install devtools if github array not empty
• determine size overhead due to devtools
• script to install from github

Problem:

• what is printed to console might get logged

TODO:

• make a list of all places where we display secrets; the most important being AWS secrets
• remove secrets from echo

Problem

• not all AWS CLI commands are checked to see if execution was successful

TODO

• add exist status check after each AWS command (original script 14)
• fail w/ message if not successful

Problem:

• This version uses Python 2.7 and rpy2==2.8.4

Solution:

• Switch to Python 3.6

TODO:

• install Python 3.6, pip3, update pip, install rp2 for python 3
• make it work / test with R 3.4.x

I think the tooling has improved considerably and its fairly straightforward to compile binaries within a docker image without having to spin up an ec2. I'm considering using this for working with some of my colleagues that work exclusively in R and was hoping to make it as easy and painless as possible for them to get an RScript scraper deployed with precompiled binaries

I used this repo to build some python library binaries in the past: https://github.com/AlJohri/aws-lambda-lxml. the whole thing boiled down to:

docker run -v $(pwd)/"$version/py36/":/outputs -it lambci/lambda:build-python3.6 pip install lxml=="4.2.4" -t /outputs/
tar -czvf "4.2.4/py36/lxml-4.2.4.tgz" "4.2.4/py36/lxml"
aws s3 cp 4.2.4/py36/lxml-4.2.4.tgz s3://mybucket/lambda-compiled-binaries/py36/

see build.sh

would you consider a PR switching to docker instead of ec2?

Great project!

I am trying to configure the setup_user.sh script file by following the example file.

I added:

#!/bin/bash

# user settings and secrets
# customize and rename to `setup_user.sh`
# overwrites `secrets_default.sh`, `settings_default.sh`, `setup_auto.sh`
PRJ_NAME ="myproj"
PRJ_BRANCH="myproj"

Then I run scripts/01_main.sh

Got error:

\e[32mINFO :\e[39m Checking project directories
\e[32mINFO :\e[39m Loading default settings and secrets
/Users/macos/Documents/r-ec2/myproj/settings/setup_user.sh: line 6: PRJ_NAME: command not found
\e[31mERROR:\e[39m PRJ_NAME is \e[95mMISSING\e[39m. Exiting.

Is there a simple way to configure the setup_user script?

Open list / wishes / references. Once fully defined, move them to separate issues

Problem

• documentation is not complete, there is setup taking place outside scripts

TODO

• Instructions for each secret
  • 2-3 lines explaining where to find the info/value, similar to code comments
  • 1 (max 2) screenshots to help visualize the issue
• Instructions for what is not in script
  • big picture first
  • most important: do not forget to include any major configuration areas
• instructions to terminate VMs using AWS web console