eclair

Erlang Config Layer. eclair is an escript that merges Erlang config stored in a well-defined hierarchy of S3 paths.

Getting started

These steps will get eclair set up to pull config from an S3 path.

make build
cp eclair.tar ~/path/to/myapp && cd ~/path/to/myapp
tar xfv eclair.tar
./eclair.erl bootstrap -version auto

Note: Your S3 credentials will be stored in .eclair/secure.config.

Uploading config

eclair is able to merge together standard Erlang proplist config files. Here's an example:

[
    {myapp, [
        {config_item, "Hi!"}
    ]}
].

We'll consider this file our root config and upload it to s3://${ECLAIR_ROOT}/myapp/root/myapp.config.

Next, our app has a config element specific to a QA environment. We'll upload the following files to s3://${ECLAIR_ROOT}/myapp/env/qa/myapp.config.

[
    {myapp, [
        {password, "myQApassword"}
    ]}
].

And PROD config to s3://${ECLAIR_ROOT}/myapp/env/pro/myapp.config.

[
    {myapp, [
        {password, "myPRODpassword"}
    ]}
].

When myapp is deployed, the deploy script must run

./eclair.erl -tags env/qa

The final config will contain this data:

[
    {myapp, [
        {config_item, "Hi!"},
        {password, "myQApassword"}
    ]}
].

Order of Operations

bootstrap

The bootstrap step is optionally run to help the user set up the necessary eclair config. Most importantly this is the s3 access_key and secret_key. It's exercisted with ./eclair.erl bootstreap, and command line arguments can be given (e.g. -version 1.0.1)

1. Read command line arguments
2. Read from .eclair/secure.config
3. Read from ~/.s3cfg (access_key and secret_key only)
4. Read from input prompt
5. Write .eclair/secure.config

main

This represents the main operation of eclair. It's exercised with ./eclair.erl, and command line arguments can be given to override anything in .eclair/secure.config.

Setup

1. Read command lien arguments
2. Read from .eclair/secure.config
3. Read from hardcoded values in eclair.erl macros

Gather data

1. Get list of nodes from epmd
2. Get hostname
3. Find application details in an ebin/*.app file

Build S3 paths

1. Append application name from *.app file to eclair root
2. List s3 common prefixes for the version subkey
3. Find existing version-root that is closest to the version input. (if none, this step is skipped)
4. Search for config with these subkeys: root, input/tag1, input/tag2, epmd/node1, epmd/node2, host/this-host

For each path, merge config

1. List all keys at this path
2. Create subdirectories as needed
3. If new file, get the file directly
4. If existing file, attempt a file:consult on both, and merge
5. If consult fails, overwrite existing file

release

The release command does everything that the main does, but will search for the existence of releases/RELEASES to aid in the search for the app's root directory. If releases/RELEASES does not exist, then the current directory of eclair.erl is considered the root.