Why I can't put proxy_set_header inside an if clause? - configuration

With this configuration:
server {
listen 8080;
location / {
if ($http_cookie ~* "mycookie") {
proxy_set_header X-Request $request;
proxy_pass http://localhost:8081;
}
}
}
I have this error when I reload nginx service:
Reloading nginx configuration: nginx: [emerg] "proxy_set_header" directive is not allowed here in /etc/nginx/conf.d/check_cookie.conf:5
nginx: configuration file /etc/nginx/nginx.conf test failed
This configuration works OK, but it does not do what I want:
server {
listen 8080;
location / {
proxy_set_header X-Request $request;
if ($http_cookie ~* "mycookie") {
proxy_pass http://localhost:8081;
}
}
}
Why I can't put proxy_set_header directive inside an if clause?

Inside location try something like this
# default header value in a new variable
set $esb "$remote_addr, $host";
# if my custom header exists
if ($http_my_header){
set $esb "$http_my_header, $remote_addr, $host";
}
proxy_set_header my-header $esb;

Unlike proxy_pass, you cannot put proxy_set_header inside an if block. You can only put it in http/server/location block. So your 2nd config is good.
Reference: http://nginx.org/en/docs/http/ngx_http_proxy_module.html#proxy_set_header
context: http, server, location
Don't know what the $request variable is. It doesn't appear in nginx variable list: http://wiki.nginx.org/HttpCoreModule#Variables. What are you trying to achieve here?

Related

Nginx | 2 Domains (1x Node App, 1x Static HTML) on one server

I am having trouble running one node app and one static page (just html) on two seperate domains at the same time. No matter what I tried the static domain gets always redirected to the node app (on port 3000)
Here are the "sites-available" files :
Node App :
server {
listen [::]:80;
listen 80;
server_name www.domain1.com domain1.com;
# and redirect to the https host (declared below)
return 301 https://domain1.com$request_uri;
}
server {
listen 443;
server_name domain1.com www.domain1.com;
ssl on;
# Use certificate and key provided by Let's Encrypt:
ssl_certificate /etc/letsencrypt/live/domain1.com/fullchain.pem;
ssl_certificate_key /etc/letsencrypt/live/domain1.com/privkey.pem;
ssl_session_timeout 5m;
ssl_protocols TLSv1 TLSv1.1 TLSv1.2;
ssl_prefer_server_ciphers on;
ssl_ciphers 'EECDH+AESGCM:EDH+AESGCM:AES256+EECDH:AES256+EDH';
location / {
proxy_set_header X-Real-IP $remote_addr;
proxy_set_header Host $http_host;
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
proxy_set_header X-NginX-Proxy true;
proxy_pass http://localhost:3000/;
proxy_ssl_session_reuse off;
proxy_set_header Host $http_host;
proxy_cache_bypass $http_upgrade;
proxy_redirect off;
}
}
And the static one :
server {
listen [::]:80;
listen 80;
#server_name www.domain2.com domain2.com;
root /var/www/html/domain2;
index index.html index.htm;
return 301 https://domain2.com$request_uri;
}
server {
listen [::]:443 ssl;
listen 443 ssl;
root /var/www/html/domain2;
index index.html index.htm;
ssl_certificate /etc/letsencrypt/live/domain2.com/fullchain.pem;
ssl_certificate_key /etc/letsencrypt/live/domain2.com/privkey.pem;
}
The default config file is empty. Any help/hint would be greatly appreciated.
It worked fine until I generated a Let's encrypt certificate for domain2, put both domains in seperate configs and removed the default.
Thank you in advance!
The problem is that you have no server_name directive in your static domain configuration. As a result, the request is always caught by your default server block, which appears to be your node app.
See for details:
How nginx processes a request
Server names
Configuring HTTPS servers

Configure Nginx to use several directories

There config nginx that when referring to http://example.com/sellers
must give the server from folder /data/sellers
In another case - of folder /data/customers
Nginx config:
server {
listen 80;
server_name localhost;
index index.html index.htm home.html;
location /sellers {
proxy_set_header X-Real-IP $remote_addr;
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
proxy_set_header Host $http_host;
proxy_set_header X-NginX-Proxy true;
rewrite ^/sellers/?(.*) /$1 break;
root /data/sellers;
}
location / {
root /data/customers;
}
}
Everything works, but is not entirely correct: when accessing the server sellers somehow uses index.html from folder /data/customers/index.html
And all the rest of the folder /data/sellers (this is correct)
What can be wrong? Why nginx takes the wrong index.html file, even though all the rest of Georgia right?
Answer
server {
listen 80;
server_name localhost;
index index.html index.htm home.html;
root /data/customers;
location /sellers {
alias /data/sellers;
}
}

Configuring nginx for single page website with HTML5 push state URL's

How can I configure nginx to redirect all URL's (not prepended with /api or some static resource eg. JS/images) to index.html? Reason is I am using HTML5 push state URL's with a single page application. Meaning content is changed whether AJAX or JS depending on the URL
My current nginx config looks like:
server {
listen 2000;
server_name localhost;
location / {
root /labs/Projects/Nodebook/public;
index index.html;
}
location /api/ {
proxy_set_header X-Real-IP $remote_addr;
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
proxy_set_header Host $http_host;
proxy_set_header X-NginX-Proxy true;
proxy_pass http://localhost:3000/;
proxy_redirect off;
}
}
location / {
try_files $uri /index.html;
}
This will check if the requested file exists and return it. If the file doesn't exist, it will return index.html.
http://nginx.org/en/docs/http/ngx_http_core_module.html#try_files
mattes answer is almost a solution, however it won't give 404 for missing files (e.g. favicon.icon) as aschepis pointed out.
Nginx will pick the first location that matches. So we can first match for files (which will give 404 if the file does not exist). And after put a location which defaults to index.html for all urls.
location /.+\..+ { # files (assuming they always have a dot)
# use eg alias to serve some files here
}
location / { # url routed by client, client gives 404 for bad urls
try_files $uri /index.html;
}
You need to add to your nginx config file:
rewrite ^(.+)$ /index.html last;
Then say you're using Backbone.js just make sure you re-route any non-defined route to a 404 page:
routes: {
// Other routes
"*path" : "notFound"
},
notFound: function(path) {
// Load 404 template, probably of a cute animal.
}
Source:
http://readystate4.com/2012/05/17/nginx-and-apache-rewrite-to-support-html5-pushstate/

nginx reverse proxy: How to implement?

I'm trying to do a reverse proxy with Nginx based on the URL. I want http://mydomain.example.com/client1/... to be redirected to http://127.0.0.1:8193/.... I have tried many ways, and none of them worked. Please note that the application can make redirections. These are the configuration files of my last solution :
default
server {
listen 80;
server_name mydomain.example.com;
location / {
set $instance none;
if ($request_uri ~ ^/(.*)/$) {
set $instance $1;
}
set $no_cookie true;
if ($http_cookie ~ "instance=([^;] +)(?:;|$)") {
set $instance $1;
set $no_cookie false;
}
if ($no_cookie = true) {
add_header Set-Cookie "instance=$instance;Domain=$host;Path=/";
rewrite ^ / break;
}
include instances.conf;
}
instances.conf
proxy_redirect off;
proxy_set_header Host $host;
proxy_set_header X-Real-IP $remote_addr;
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
proxy_connect_timeout 90;
proxy_send_timeout 60;
# Installation of language packs, etc. can take a long time
proxy_read_timeout 10800;
if ($instance = client1) {
proxy_pass http://127.0.0.1:8193;
}
if ($instance = client2) {
proxy_pass http://127.0.0.1:8194
}
...
When the browser requests http://mydomain.example.com/client1/, Nginx should set a cookie named instance with the value client1 then redirect the traffic to the appropriate proxy. For subsequent queries, it should use this cookie to make redirection. The problem I have is it never sets the $instance variable to client1. Don't forget that the application has no idea of the prefix /client1.
Do you have an idea? Do you know of a better solution?
The regex used to get the cookie was wrong. I have changed this to
"instance=([^;][^ ]+)(?:;|$)"
and it works now.
Edit: It's only a part of the solution finally. I'm sorry. There is still a problem. See my comment below.
It is not related to your problem but "proxy_connect_timeout"
"This directive assigns a timeout for the connection to the upstream server. It is necessary to keep in mind that this time out cannot be more than 75 seconds."
See Nginx' map module
map $uri $proxy {
/client1 http://127.0.0.1:8193/client1;
/client2 http://127.0.0.1:8194/client2;
}
server {
server_name my.domain.com;
proxy_pass $proxy;
}
Note that appending /clientX to the end of the proxy_pass URI strips that portion of the URI from the request (which seems rational to me, but may not be what you want).

Nginx to server node.js content + static content

I have an application with both dynamic and static content. I use nginx as a front end for this app. When dynamic content is requested, the request is forwarded to an unix socket (to a node.js app), this part works well. I have added a "location" directive to serve the static content but this part does not work, I get the 404 error each time despite the fact the folder "/home/test/my_app/static" does exist.
This is the nginx conf I have:
upstream test_sock {
server unix:/tmp/test.sock
fail_timeout=0;
}
server {
listen 15000;
client_max_body_size 4G;
server_name localhost domain.com;
keepalive_timeout 5;
location ~ /static/ {
if (!-f $request_filename) {
return 404;
}
if (-f $request_filename) {
root /home/test/my_app/static;
expires 30d;
}
}
location / {
proxy_pass http://test_sock;
proxy_redirect off;
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
proxy_set_header Host $http_host;
}
}
Any idea ?
hmmm... ok, silly thing, I was missing the root directive before the locations ones...