I want to run a shell script from my Rails application. The script is started on the remote server via the net-ssh
Gem. Here is my code:
Net::SSH.start('localhost', 'user', :password => '1234', :port => port) do |session|
session.exec!("#{Rails.root}/script/myscript")
end
I have checked that the script is present in my local application. Script is taking about 1 hour for completion. I have two questions:
The sample doc says that the simple, and quite proper way to run the Net::SSH
session is the following:
HOST = '192.168.1.113'
USER = 'username'
PASS = 'password'
Net::SSH.start( HOST, USER, :password => PASS ) do| ssh |
result = ssh.exec! 'ls'
puts result
end
I recomment to pass at least password argument via shell environment to don't store it in the script plainly. Also you could use micro-optparse
gem to pass other argument via command line. So it could be as follows:
require 'micro-optparse'
options = Parser.new do |p|
p.banner = "This is a fancy script, for usage see below"
p.version = "fancy script 0.0 alpha"
p.option :host, "set host", :default => 'localhost'
p.option :host, "set user", :default => ''
end.parse!
Net::SSH.start( options[ :host ], options[ :user ], :password => ENV[ 'PASSWORD' ] ) do| ssh |
result = ssh.exec! 'ls'
puts result
end
And run from command line:
$ bundle exec ./1.rb --host=111.ru --user=user
{:host=>"111.ru", :user=>"user"}
Of course the support for :port
argument can be added in the same manner.
Use nohup
or screen
utitilies to run a script as a service in linux:
result = ssh.exec! "nohup #{Rails.root}/script/myscript"