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

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.

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"}
    ]}
].

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

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.

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

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

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

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

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.