in progress
All of the code in the repo was originally written to run locally on OSX. I am now porting most of it to run on Amazon Web Services (AWS).
Log into your AWS console (I'm using the Stamen group). I'm using the us-west-2a zone (Oregon).
Create a 100GB gp2 EBS volume for the history dump. (As for Jan 2016 the history dump is 49G itself)
Create an r3.xlarge ubuntu instance to do the history splitting (and processing?). This is a pretty big instance, so remember to stop it when its done!
Log into the instance:
ssh -i ~/.ssh/osmhistory.pem [email protected]
or
ssh -i ~/.ssh/osmhistory.pem [email protected]
(get latest IP address from the EC2 console)
Associate the EBS volume with the instance using the web console. Then mount it:
List the disks:
sudo fdisk -l
Make the filesystem:
sudo mkfs -t ext4 /dev/xvdf
Make a mountpoint:
sudo mkdir /mnt/ebs
Mount the drive:
sudo mount /dev/xvdf /mnt/ebs
Confirm that it mounted:
mount -l
df
Download the history file in pbf format. Find the link to the latest version at the bottom of this page: http://planet.openstreetmap.org/planet/experimental/
wget http://planet.openstreetmap.org/planet/experimental/history-2014-11-24.osm.pbf
Then create a symlink to this file using .osh.pbf instead of osm.pbf (necessary for the osm-history-splitter to read it correctly)
ln -s history-2014-11-24.osm.pbf history-2014-11-24.osh.pbf
Then make a snapshot of this volume.
After you run the osm-history-splitter you probably want to make another snapshot of this volume.
After you're finished doing the splitting, you can unmount this drive and stop this large instance. Then connect this drive (the one with the extracts) to a smaller EC2 instance to import them into a database (using osm-history-renderer) and run the map-gardening analysis.