Skip to content

A hobby project to do cloud processing with AWS. There's almost for sure a better way.

Notifications You must be signed in to change notification settings

gadomski/magic-bucket

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

41 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Magic bucket

An AWS S3 bucket that uses a Docker container to process pointcloud files. Notifications from the magic bucket are posted to the #magic-bucket channel in our Slack.

Usage

The S3 container is s3://crrel-magic-bucket. Top-level directories in the bucket define the tasks that are available. Each task has its own workflow and data organization expectations, described below.

Task: pdal-translate

Usage

aws s3 cp config.json s3://crrel-magic-bucket/pdal-translate/my-task/config.json
aws s3 cp infile.las s3://crrel-magic-bucket/pdal-translate/my-task/infile.las
# wait for the task to complete
aws s3 cp s3://crrel-magic-bucket/pdal-translate/my-task/output/infile.las outfile.las

The magic bucket will download infile.las, read configuration from config.json, and run everything through pdal translate. The output file will be uploaded to s3://crrel-magic-bucket/pdal-translate/my-task/output/infile.las.

Detail

The pdal-translate task takes an input file and produces and output file, optionally applying filters and other modifications along the way. The input file can be uploaded one of two ways:

  • As a point cloud file. e.g. a file with an .las extension.
  • As a gzipped (.gz) or zipped (.zip) archive, in which case the archive will be expanded before processing.

The configuration file is JSON. pdal-translate looks for a configuration file in the following order:

  1. If you include a config.json in your gzipped or zipped archive, that configuration file will be used.
  2. If the input file is s3://crrel-magic-bucket/pdal-translate/to-laz/simple.las, the configuration file is named s3://crrel-magic-bucket/pdal-translate/to-laz/simple.las.json.
  3. If the input file is s3://crrel-magic-bucket/pdal-translate/to-laz/simple.las, the configuration file is named s3://crrel-magic-bucket/pdal-translate/to-laz/config.json.

Configuration files look like this (all fields optional):

{
  "filters": [], // a list of filters to be applied by `pdal translate`
  "output_ext": ".laz", // the extension of the output file (used to specify format)
  "args": ["--writers.las.scale_x", ".1"] // additional arguments to pass to `pdal translate`
}

See the examples directory in this repo for sample configuration files.

Task: pdal-info

Runs pdal info --all on a file:

aws s3 cp autzen.las s3://crrel-magic-bucket/pdal-info/autzen.las
aws s3 cp s3://crrel-magic-bucket/pdal-info/output/autzen.las.json autzen.las.json

Task: rimtatls

Runs Riegl's rimtatls executable on rxp files:

aws s3 cp 160520_181202.rxp s3://crrel-magic-bucket/rimtatls/160520_181202.rxp
aws s3 cp s3://crrel-magic-bucket/rimtatls/output/160520_181202.mta.rxp 160520_181202.mta.rxp

An .mta is inserted into the filename to indicate that it has been MTA processed.

Architecture

The magic bucket consists of these parts:

  • An S3 bucket called crrel-magic-bucket.
  • An AWS lambda function, called magic-bucket, whose code lives in lambda.py. This function takes an S3 event and does two things:
    1. Puts a message into an SQS queue, called magic-bucket, containing the contents of the S3 event.
    2. Starts an ECS task, called magic-bucket, to process all messages in SQS queue.
  • An ECS task, called magic-bucket, that runs a Docker container, called magic-bucket. The docker container is built with docker/Dockerfile and runs the code in docker/main.py.

About

A hobby project to do cloud processing with AWS. There's almost for sure a better way.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages