MAIL_URL does not work on *.meteor.com as described in the documentation

[landland]

The documentation here states that the default Mailgun MAIL_URL may be overridden by setting process.env.MAIL_URL, however this does not appear to be the case. The documentation states:

For apps deployed with meteor deploy, MAIL_URL defaults to an account (provided by Mailgun) which allows apps to send up to 200 emails per day; you may override this default by assigning to process.env.MAIL_URL before your first call to Email.send.

However, Mailgun is always used despite the process.env.MAIL_URL setting. To reproduce:

1. create new meteor app

2. add below js code

3. modify code to specific smtp provider (such as from Mandrill) and appropriate from and to email addresses

4. deploy to a *.meteor.com domain using meteor deploy

5. visit the domain and click the button

6. verify that email is from Mailgun and not the smtp provider specified in process.env.MAIL_URL

   if (Meteor.isClient) {
     Template.hello.greeting = function () {
       return "Welcome to testmail.";
     };
     Template.hello.events({
       'click input' : function () {    
         Meteor.call('sendEmail',
               'xxx@gmail.com',
               'xxx@domain.com',
               'Hello from Meteor!',
               'This is a test of Email.send.');           
       }
     });
   }
   
   if (Meteor.isServer) {
   Meteor.methods({
     sendEmail: function (to, from, subject, text) {
       check([to, from, subject, text], [String]);
       this.unblock();
       Email.send({
         to: to,
         from: from,
         subject: subject,
         text: text
       });
     }
   });
     Meteor.startup(function () {    
       process.env.MAIL_URL = 'smtp://user:password@smtp.mandrillapp.com:587/';          
    });
   }

If this behavior is intended, then please update documentation to reflect this.

See Stack Overflow thread http://stackoverflow.com/questions/20337309/meteor-deploy-mail-url-not-being-set for more details

[landland]

Ah, looks like Tarangp reported this earlier today here #1647

[glasser]

Thanks. I'm taking a different approach to @Tarangp that fixes some related issues, so I'll keep this one open and close the PR.