diff --git a/.github/workflows/demo.yml b/.github/workflows/demo.yml
index 0d860f1..c7c7511 100644
--- a/.github/workflows/demo.yml
+++ b/.github/workflows/demo.yml
@@ -1,7 +1,21 @@
on: [push, pull_request]
jobs:
- load_key_demo:
+ single_key_demo:
+ strategy:
+ matrix:
+ os: [ubuntu-latest, macOS-latest]
+ runs-on: ${{ matrix.os }}
+ steps:
+ - uses: actions/checkout@v1
+ - name: Setup key
+ uses: ./
+ with:
+ ssh-private-key: |
+ ${{ secrets.DEMO_KEY }}
+ ${{ secrets.DEMO_KEY_2 }}
+
+ multiple_keys_demo:
strategy:
matrix:
os: [ubuntu-latest, macOS-latest]
@@ -12,8 +26,6 @@ jobs:
uses: ./
with:
ssh-private-key: ${{ secrets.DEMO_KEY }}
- - run: |
- ssh-add -l
- echo SSH_AUTH_SOCK is at $SSH_AUTH_SOCK
-
+
+
diff --git a/CHANGELOG.md b/CHANGELOG.md
new file mode 100644
index 0000000..7d07023
--- /dev/null
+++ b/CHANGELOG.md
@@ -0,0 +1,14 @@
+# Changelog
+
+All notable changes to this project will be documented in this file.
+
+The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
+and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
+
+## [Unreleased]
+
+### Added
+
+* Multiple SSH keys can now be provided (#14, closes #7). Thanks to
+ @webknjaz and @bradmartin for support and tests.
+
diff --git a/README.md b/README.md
index 59c90e7..81484bb 100644
--- a/README.md
+++ b/README.md
@@ -15,10 +15,10 @@ GitHub Actions only have access to the repository they run for. So, in order to
## Usage
1. Create an SSH key with sufficient access privileges. For security reasons, don't use your personal SSH key but set up a dedicated one for use in GitHub Actions. See below for a few hints if you are unsure about this step.
-2. In your repository, go to the *Settings > Secrets* menu and create a new secret called `SSH_PRIVATE_KEY`. Put the *unencrypted private* SSH key in `PEM` format into the contents field.
- This key should start with `-----BEGIN RSA PRIVATE KEY-----`, consist of many lines and ends with `-----END RSA PRIVATE KEY-----`.
- You can just copy the key as-is from the private key file.
-3. In your workflow definition file, add the following step. Preferably this would be rather on top, near the `actions/checkout@v1` line.
+2. Make sure you don't have a passphrase set on the private key.
+3. In your repository, go to the *Settings > Secrets* menu and create a new secret. In this example, we'll call it `SSH_PRIVATE_KEY`. Put the contents of the *private* SSH key file into the contents field.
+ This key should start with `-----BEGIN ... PRIVATE KEY-----`, consist of many lines and ends with `-----END ... PRIVATE KEY-----`.
+4. In your workflow definition file, add the following step. Preferably this would be rather on top, near the `actions/checkout@v1` line.
```yaml
# .github/workflows/my-workflow.yml
@@ -34,7 +34,28 @@ jobs:
ssh-private-key: ${{ secrets.SSH_PRIVATE_KEY }}
- ... other steps
```
-4. If, for some reason, you need to change the location of the SSH agent socket, you can use the `ssh-auth-sock` input to provide a path.
+5. If, for some reason, you need to change the location of the SSH agent socket, you can use the `ssh-auth-sock` input to provide a path.
+
+### Using multiple keys
+
+There are cases where you might need to use multiple keys. For example, "deployment keys" might be limited to a single repository each.
+
+In that case, you can set-up the different keys as multiple secrets and pass them all to the action like so:
+
+```yaml
+# ... contens as before
+ - uses: webfactory/ssh-agent@v0.1.1
+ with:
+ ssh-private-key: |
+ ${{ secrets.FIRST_KEY }}
+ ${{ secrets.NEXT_KEY }}
+ ${{ secrets.ANOTHER_KEY }}
+```
+
+The `ssh-agent` will load all of the keys and try each one in order when establishing SSH connections.
+
+There's one **caveat**, though: SSH servers may abort the connection attempt after a number of mismatching keys have been presented. So if, for example, you have
+six different keys loaded into the `ssh-agent`, but the server aborts after five unknown keys, the last key (which might be the right one) will never even be tried.
## Known issues and limitations
@@ -72,11 +93,12 @@ As a side note, using `ssh-keyscan` without proper key verification is susceptib
## Creating SSH keys
-In order to create a new SSH key, run `ssh-keygen -t rsa -b 4096 -m pem -f path/to/keyfile`. This will prompt you for a key passphrase and save the key in `path/to/keyfile`.
+In order to create a new SSH key, run `ssh-keygen -t ed25519 -a 100 -f path/to/keyfile`, as suggested in [this blog post](https://stribika.github.io/2015/01/04/secure-secure-shell.html).
+If you need to work with some older server software and need RSA keys, tr `ssh-keygen -t rsa -b 4096 -o -f path/to/keyfile` instead.
-Having a passphrase is a good thing, since it will keep the key encrypted on your disk. When configuring the secret `SSH_PRIVATE_KEY` value in your repository, however, you will need the private key *unencrypted*.
-
-To show the private key unencrypted, run `openssl rsa -in path/to/key -outform pem`.
+Both commands will prompt you for a key passphrase and save the key in `path/to/keyfile`.
+In general, having a passphrase is a good thing, since it will keep the key encrypted on your disk. When using the key with this action, however, you need to make sure you don't
+specify a passphrase: The key must be usable without reading the passphrase from input. Since the key itself is stored using GitHub's "Secret" feature, it should be fairly safe anyway.
## Authorizing a key
@@ -93,7 +115,7 @@ As a note to my future self, in order to work on this repo:
* Clone it
* Run `npm install` to fetch dependencies
* _hack hack hack_
-* `node index.js` (inputs are passed through `INPUT_` env vars, but how to set `ssh-private-key`?)
+* `node index.js`. Inputs are passed through `INPUT_` env vars with their names uppercased. Use `env "INPUT_SSH-PRIVATE-KEY=\`cat file\`" node index.js` for this action.
* Run `./node_modules/.bin/ncc build index.js` to update `dist/index.js`, which is the file actually run
* Read https://help.github.com/en/articles/creating-a-javascript-action if unsure.
* Maybe update the README example when publishing a new version.
diff --git a/dist/index.js b/dist/index.js
index 2ae7540..602df64 100644
--- a/dist/index.js
+++ b/dist/index.js
@@ -72,7 +72,11 @@ try {
core.exportVariable('SSH_AUTH_SOCK', authSock);
console.log("Adding private key to agent");
- child_process.execSync('ssh-add -', { input: core.getInput('ssh-private-key') });
+ core.getInput('ssh-private-key').split(/(?=-----BEGIN)/).forEach(function(key) {
+ child_process.execSync('ssh-add -', { input: key.trim() + "\n" });
+ });
+ console.log("Keys added:");
+ child_process.execSync('ssh-add -l', { stdio: 'inherit' });
} catch (error) {
core.setFailed(error.message);
}
diff --git a/index.js b/index.js
index 141ad08..140afb4 100644
--- a/index.js
+++ b/index.js
@@ -17,7 +17,11 @@ try {
core.exportVariable('SSH_AUTH_SOCK', authSock);
console.log("Adding private key to agent");
- child_process.execSync('ssh-add -', { input: core.getInput('ssh-private-key') });
+ core.getInput('ssh-private-key').split(/(?=-----BEGIN)/).forEach(function(key) {
+ child_process.execSync('ssh-add -', { input: key.trim() + "\n" });
+ });
+ console.log("Keys added:");
+ child_process.execSync('ssh-add -l', { stdio: 'inherit' });
} catch (error) {
core.setFailed(error.message);
}