Simple parallel server cluster management tool for Node using ssh and scp. Specify an ssh or scp command and it will run on all servers in the cluster in parallel.
npm install shync
var Shync = require('shync').Shync;
var cluster = new Shync({
domains: ['23.42.103.164',
'mydomain.com'],
user: 'ubuntu',
keyLoc: '/path/to/public/key',
bypassFingerprint: true
});
cluster.run('node ~/stuff/important.js', function(err){
if (err) return console.log(err);
console.log('you\'ve done something important on many machines!');
});cluster.run('~/important.tar.gz', '~/stuff/important.tar.gz', function(err){
if (err) return console.log(err);
console.log('you\'ve done something important on many machines!');
});cluster.run('~/important.tar.gz', '~/stuff/important.tar.gz', function(err){
if (err) return console.log(err);
cluster.run('tar xzvf ~/stuff/important.tar.gz', function(err){
if (err) return console.log(err);
cluster.run('node ~/stuff/important.js', function(err){
if (err) return console.log(err);
console.log('you\'ve done something important on many machines!');
}
});
}
});
}
});If a command was not successful, shync callbacks will receive the return code of the command that was run as the err paramater. shync treats the success of a command as "all or nothing." If the command was successful on all remote machines, null is returned to the callback. If even one machine returned a non 0 code, then the command as a whole is deemed to have failed and the non 0 code will be returned as the err param to the callback.
shync fails quickly if a remote machine returns a non 0 code. All outstanding connections to the remaining servers are severed, since their return codes are irrelevant now that a non 0 code has been received.
You can specify a function to be called with the stdout and/or stderr of the program you're running:
function stdout(out) {
console.log(out);
}
function stderr(err) {
console.log(err);
}
var cluster = new Shync({
...
stdout: stdout,
stderr: stderr
});
cluster.run('rm -rf /', cb);
// ec2-54-226-122-165.compute-1.amazonaws.com:STDERR: rm: it is dangerous to operate recursively on `/'
// ec2-54-226-122-165.compute-1.amazonaws.com:STDERR: rm: use --no-preserve-root to override this failsafe
cluster.run('ls -a ~', cb);
// ec2-54-226-122-165.compute-1.amazonaws.com:STDOUT: . .. .bash_history .bashrc .sshvar cluster = new Shync({
...
bypassFingerprint: true
...
});You know when you ssh from the command line into a brand new box, and you get this message:
$ ssh user@example.com
The authenticity of host 'example.com (23.20.102.179)' can't be established.
RSA key fingerprint is 2f:5d:69:26:e2:a0:23:53:f7:0a:21:51:0a:74:8a:49.
Are you sure you want to continue connecting (yes/no)?
bypassFingerprint: true stops this from happening.
Warning Use bypassFingerprint at your own risk.