I have a Rails (2.3.8) application that will need to start and maintain a SSH tunnel whenever the application is started using 'script/server' or when started using Passenger. When script/server is ^C'd or the Passenger instance is shut down the SSH tunnel should be destroyed.
I do not want the tunnel to be started when I run 'script/console' - so using config/environment.rb doesn't look like a good option.
As such, I don't want the tunnel to be backgrounded - I want it to be attached and owned by the ruby process, and I only want one tunnel per server.
The tunnel itself will most likely be started by running SSH directly, but if there's a simple way to do it using the SSH libraries I'll use that instead.
Is there a way to do this in Rails? I can think of a way to do it using config/environment.rb using a series of lock files and other bits of messing around, but I was hoping for some sort of :on_server_start and :on_server_exit hook.
--
For the curious, I need to do this as the Rails application is running in location A and will receive updates from a series of services in location B. Location B, however, does not have a direct route to location A. I will be starting a SSH tunnel from the application to a machine in location B and the services in location B will send updates to that machine instead.
Managed to solve this one using config/environments/*.rb and a quick fork hack. See gist: http://gist.github.com/466267
So, it'll be run if you use RAILS_ENV=production but not for test/development.
Related
Currently, we are running our rails app, on AWS but we tried to move it to Heroku, which didn't work at all, now We are trying to run the app on Virtual Machine on hetzner. I can run the app on local server easily, my question is that, How can we run our rails app with a specific IP and then we can access that app anywhere in the world from that ip? is that possible to do so? We are using PUMA for server.
Yes, you can access to to your site with specific IP.
In description of hetzner Virtual Server says that server have 1 specific IP. So than you deployed and run you application y can access to app with 1.1.1.1:80 or another port.
If i understand you. Also what happened with heroku?
I have two questions about rails server:
Do I have to start the server from within the application folder?
Is the server I started only for that application?
If they are true, this does not quite make sense to me, since why do I need to start multiple servers?
Or is there some kind of master configuration, so that one server can route to different applications? Is Capistrano for this purpose?
I'm going to assume you're talking about the rails server command, for running a local rails server to test your application, and that you're not talking about setting up a rails application on a remote server. Please mention if that is not the case.
Yes, you must execute rails server from within the root folder of your rails application.
Yes, the server you started is only for that application. It's a self-contained thing.
You should not need to start multiple servers. Even if you have multiple applications, you probably don't need to have more than one running at a time. So, you can shut down the rails server in one application (Ctrl-C) and then cd to your new application, and start a new rails server there with rails server.
If you do need to run two local rails applications at once, you can do so by running them on different ports. So, the first one, you can just execute rails server and it will make your site available at localhost:3000 (because port 3000 is the default port). The next one, you can specify a port other than 3000 - eg. rails server -p 3001 to get a rails app at localhost:3001.
Capistrano is for deploying your applications to a remote server, not for running them locally on your own computer. So, it is not relevant here. What you may be interested in is http://pow.cx/
Again, I've assumed you're talking about running your rails app locally on your own computer. If you're referring to deploying it to the internet on a server, then you can ignore this answer.
Problem:
My rails app (on my local machine) only responds to requests sent from the same machine to localhost, 127.0.0.1, or my internal ip address. When I try to hit it using my internet ip or from any other machine, inside or outside of my network, it just times out. I'm on Mac OS 10.9.1, ruby 1.9.3, rails 4.0.0.
I've done a lot of searching but all I can find is problems where people didn't forward their ports or bind the right ip.
Here are the areas I've investigated:
Ports -
I've tried several different ports. I configured my router to forward every port I tried but got the same result. I thought maybe there was a problem with the router so I built a simple server in Java and bound all the same ports I was binding with my rails app. Sure enough, when I hit the Java app using my internet ip it worked just fine so the router/firewall/port forwarding isn't the problem. Also, I run an apache server on port 80 and that has never had any problems. I turned apache off and tried port 80 for my rails app but that didn't fix the problem.
Rails Server -
I started with WEBrick and I thought that perhaps there was some setting inside that blocked external requests. I searched google extensively and found nothing on that matter. Just to be safe I installed Thin and got the exact same result I did with WEBrick. One interesting thing is that when the rails server is started, the external request takes a long time to time-out, but the server console displays no output at all. However if I try to send the same request w/out starting the server at all it fails immediately.
User Permissions -
I started the server with root (i'm starting to just shoot in the dark here) and it had no effect.
Environment -
I was starting in development environment originally because I'm developing but just for fun I tried starting in production and it also made no difference.
PLEASE HELP ME SMART PEOPLE
Update:
I installed the app on my Ubuntu machine and it doesn't have this problem! So that suggests the problem may have something to do with Mac OS.
SOLVED:
It turns out that in the System Preferences -> Security & Privacy -> Firewall in Mac OS, it was somehow set to block incoming connections to Ruby 1.9.3. I must have accidentally set that some time ago.
The problem is you are probably trying to request the page from your local machine (or any computer on your local network, behind your firewall) to your public IP expecting a result... not unless you setup routes through your firewall for this (and not usually available on a consumer level router... linksys, dlink, etc)
So forward port 80 if you are using something like pow, or 3000 for web bricks default port to your local machine
Then have someone outside your local lan request your external (public) IP
This may be related: Rails 3.1 on Ubuntu 11.10 under VirtualBox very slow
Your mention of slowness combined with the use of webrick makes me think you've got some reverse-DNS lookup awfulness going on. A quick first step is hacking /etc/hosts to bypass this lookup.
The situation I dealt with on Ubuntu was solved in the short-term by hacking /etc/hosts. You could do this quick hack in order to see if it is indeed just webrick's reverse-DNS lookup. Edit /etc/hosts and add a line for the external user's IP address, something like this:
156.123.48.55 TestPerson
Replace the IP address with the tester's IP address. Since you said you can get the external request to hit an Apache server on port 80, you can grab their IP address from the Apache access logs if necessary, otherwise just ask the person testing.
You could also try a different web server, such as unicorn, which may help out. Add "gem unicorn-rails" to your Gemfile, run bundle install, and then (according to their docs), rails server will just use unicorn directly.
With any local server, you'll need to correctly configure port forwarding on your firewall. Like said by CaptChrisD, tests must be done by an external IP/browser (if you own a server, ssh on it, then w3m to test).
I already had same symptom (server started => timeout, server stopped => fail) and the origin was an issue with firewall configuration. I think it is your problem.
With MacOS, Pow is really awesome: installation is easy, no configuration required (no /etc/hosts…). Moreover, they give you a hook for external access to your virtualhosts (but you still need port forwarding on your firewall).
Otherwise, there is other solutions like Forward to do it without firewall configuration (30-days free trial).
Hope this helps!
I am using Aptana Studio 3 for development of ROR apps. I used run server command and it showed you can access your app on {http//0.0.0.0:3000/}, but when I try to access this URL, it tells me to check your Internet connection. I tried several other ports also but it is not working. I have created/modified the files necessary and migrated the database successfully too. Appreciate any help in running the app over the browser. I am currently using WeBrick Server.
so, in your title you say "on server". what does that mean? when you are running it on a different machine than your own, you need to use the address of that machine or it's domain name. keep in mind that firewall rules might prevent any connection to that server.
when you are ON the machine, via ssh for example, you can try calling the then "local" rails instance with curl http://localhost:3000/ to verify that it is running.
I'm developing an small application on RoR and OSX 10.6.4 workstation, and I'm looking for guidance on two things: -
How to share my application locally with others so others in my team can get access to my local webserver (Mongrel?) to view/play with my system before I release it into production. From my default installation I can play with it both others can't seem to access it - not even from say http://:3000, which works from local - so I'm a bit confused.
Advice on how best to deploy it onto a production webserver assuming I deploy onto Linux. What webserver should I use and are there instructions?
Thanks in advance. Networking and webservers are NOT a strong suite :-)
Cheers
B
The development Mongrel is by default only accessible through localhost. In order to be visible from the outside you need to ask it to bind to your external IP address. Assuming your IP address is 10.0.0.5, you need to do this:
script/server -b 10.0.0.5
For deployment on Linux, the easiest way is perhaps using Passenger, and either Apache or nginx for a web server, whichever you're most comfortable with (if you have no preference, go with Apache). Documentation can be found here.
Why can't others access it via http://your.ip:3000/? ./script/server mongrel listens on 0.0.0.0 by default. 0.0.0.0 stands for listening on every interface in Linux.