Deploying Go web applications with Apache

Question

I can not find a mod_go for deploying Go web applications. Is there any other way to run web applications in Go with an Apache web server (or even IIS)?

Update: Now after doing Go full time for nearly a year; doing this (Go with Apache) is nullifying the very purpose of Go (Performance in terms of being highly concurrent). And I'm using nginx as a reverse proxy for http/https and having my Go backends behind it nicely. Though you have to change your mindset on webapps a bit when using Go.

Those interested in this can also see [a similar question](http://stackoverflow.com/q/17440415/165293). — Mostafa, Sep 09 '13 at 04:49

score 30 · Accepted Answer · edited Oct 05 '13 at 18:48

There's no mod_go. (At least I've not heard of such a thing.)

A go web app by itself is a capable web server. You can listen to port 80 in your app and then directly run it on your server. Trust me: it really works!

But if you are not doing that (for reasons such as having other virtual servers on the same machine, load balancing, etc.), you can use a HTTP server such as nginx or Apache as a HTTP proxy in front of your Go app. I use nginx and it's great. Here's an outdated but still very useful guide on how to do that with nginx. I haven't done it with Apache, but this should help.

I recommend your Go web app by itself or nginx as a HTTP proxy.

I found this mod_go on a github repo, it may be interesting to check : https://github.com/idaunis/mod_go — Mika Andrianarijaona, Jul 22 '15 at 12:56

Evan Shaw · Answer 2 · 2012-04-07T08:22:28.303

24

In addition to the other options, there's also the net/http/fcgi package. This is similar to the CGI option, but it uses FastCGI and your application can keep state if it needs to.

Here's the FastCGI version of jimt's example. Note that only two lines are different. Depending on how you configure Apache, you may have to change the first argument to something different, but nil is the common case.

package main

import (
    "fmt"
    "net/http"
    "net/http/fcgi"
)

func hello(w http.ResponseWriter, r *http.Request) {
    fmt.Fprintf(w, "Hello from Go!")
}

func main() {
    http.HandleFunc("/", hello)
    fcgi.Serve(nil, nil)
}

edited Apr 07 '12 at 08:22

answered Apr 06 '12 at 10:05

Evan Shaw

23,839
7
70
61

Do you know of any examples using net/http/fcgi? – BurntSushi5 Apr 06 '12 at 22:32
@BurntSushi5 Edited to add an example – Evan Shaw Apr 07 '12 at 08:22
There is one limitation here, however, which may be critical for someone. Quoting the package doc: "Currently only the responder role is supported". This means that http auth can't be handled by your Go code this way. – abbot Apr 09 '12 at 04:49
1

@abbot That's not correct. The authorizer FCGI role is not related to HTTP authorization. An authorizer application decides, based on the entire HTTP request, whether or not the client should have access. Then the server responds to the request if the client has access. – Evan Shaw Apr 09 '12 at 05:08

score 15 · Answer 3 · answered Dec 08 '12 at 01:16

15

For those interested there is a recently released mod_go for Apache in here:

https://github.com/idaunis/mod_go

answered Dec 08 '12 at 01:16

user1784334

191
1
4

jimt · Answer 4 · 2012-04-05T23:09:09.920

While not ideal, you can run Go programs as CGI scripts by placing them in the cgi-bin directory. You can invoke them like any other page through server.com/cgi-bin/myapp?foo=bar

An example program would look like this:

package main

import (
    "fmt"
    "net/http"
    "net/http/cgi"
)

func hello(w http.ResponseWriter, r *http.Request) {
    fmt.Fprintf(w, "Hello from Go!")
}

func main() {
    http.HandleFunc("/", hello)
    cgi.Serve(nil)
}

The reason this is not as optimal as running the program as its own server, is that with a cgi approach, the program is invoked for every single request. So any state inside it does not persist.

For clarity: You should place the compiled binary in the cgi-bin directory. Not the program source.

score 4 · Answer 5 · answered Feb 27 '19 at 03:15

This would proxy the domain and subdomain requests to the specified ports.

NameVirtualHost *:80
<VirtualHost *:80>
    ServerAdmin webmaster@dummy-host.example.com
    ServerName www.yourdomain.com
    ProxyRequests Off
    <Proxy *>
        Order deny,allow
        Allow from all
    </Proxy>
    ProxyPass / http://127.0.0.1:8080/
    ProxyPassReverse / http://127.0.0.1:8080/
</VirtualHost>

<VirtualHost *:80>
    ServerAdmin webmaster@dummy-host.example.com
    ServerName subdomain.yourdomain.com
    ProxyRequests Off
    <Proxy *>
        Order deny,allow
        Allow from all
    </Proxy>
    ProxyPass / http://127.0.0.1:8081/
    ProxyPassReverse / http://127.0.0.1:8081/
</VirtualHost>

score 2 · Answer 6 · answered Apr 09 '12 at 04:46

2

I just use the web server's proxy facility and run my app as a regular daemon (using daemonize) on the server. On apache that would be ProxyPass + ProxyPreserveHost.

answered Apr 09 '12 at 04:46

abbot

27,408
6
54
57

did you have any example? – slier Dec 06 '17 at 07:42

Deploying Go web applications with Apache

6 Answers6

Linked