Limit HTTP verbs without redundant config - nginx

I've got an Elasticsearch cluster plus Logstash and Kibana, and I only want to expose a read-only window into the indexes, with the exception of the index kibana-int so that dashboards can be saved.
I've found a suitable ES proxy config, and I've modified it to use limit_except to disallow write/modify to other indexes, but much of the config is needlessly duplicated. Is there a cleaner way to define this?
upstream elasticsearch {
server es-01.iad.company.com:9200;
server es-02.iad.company.com:9200;
}
server {
listen 9200;
server_name elasticsearch.proxy;
client_max_body_size 50m;
location / {
limit_except GET POST HEAD OPTIONS {
deny all;
}
proxy_pass http://elasticsearch;
proxy_redirect off;
proxy_set_header Connection "";
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_pass_header Access-Control-Allow-Origin;
proxy_pass_header Access-Control-Allow-Methods;
proxy_hide_header Access-Control-Allow-Headers;
add_header Access-Control-Allow-Headers 'X-Requested-With, Content-Type';
add_header Access-Control-Allow-Credentials true;
}
location /kibana-int/ {
proxy_pass http://elasticsearch;
proxy_redirect off;
proxy_set_header Connection "";
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_pass_header Access-Control-Allow-Origin;
proxy_pass_header Access-Control-Allow-Methods;
proxy_hide_header Access-Control-Allow-Headers;
add_header Access-Control-Allow-Headers 'X-Requested-With, Content-Type';
add_header Access-Control-Allow-Credentials true;
}
}

There are several ways:
Solution 1
You could put repeating config into file and include it.
Your config:
upstream elasticsearch {
server es-01.iad.company.com:9200;
server es-02.iad.company.com:9200;
}
server {
listen 9200;
server_name elasticsearch.proxy;
client_max_body_size 50m;
location / {
limit_except GET POST HEAD OPTIONS {
deny all;
}
include proxy.inc;
}
location /kibana-int/ {
include proxy.inc;
}
}
proxy.inc:
proxy_pass http://elasticsearch;
proxy_redirect off;
proxy_set_header Connection "";
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_hide_header Access-Control-Allow-Headers;
add_header Access-Control-Allow-Headers 'X-Requested-With, Content-Type';
add_header Access-Control-Allow-Credentials true;
Solution 2
Other way is use nginx's directive inheritance.
upstream elasticsearch {
server es-01.iad.company.com:9200;
server es-02.iad.company.com:9200;
}
server {
listen 9200;
server_name elasticsearch.proxy;
client_max_body_size 50m;
proxy_redirect off;
proxy_set_header Connection "";
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_hide_header Access-Control-Allow-Headers;
add_header Access-Control-Allow-Headers 'X-Requested-With, Content-Type';
add_header Access-Control-Allow-Credentials true;
location / {
limit_except GET POST HEAD OPTIONS {
deny all;
}
proxy_pass http://elasticsearch;
}
location /kibana-int/ {
proxy_pass http://elasticsearch;
}
}
BTW, your proxy_pass_header directives are needless. Nginx proxies almost all headers by default.

Related

How can I optimize my nginx configuration

I have nginx .conf file as below; I'm wondering if it can be write simpler:
default.conf
upstream docsapp {
server app:8000;
}
server {
listen 80;
location / {
alias /usr/share/nginx/html/;
}
location /admin/ {
proxy_pass http://docsapp/admin/;
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
proxy_set_header Host $host;
proxy_redirect off;
}
location /api/ {
proxy_pass http://docsapp/api/;
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
proxy_set_header Host $host;
proxy_redirect off;
}
location /api-token-auth/ {
proxy_pass http://docsapp/api-token-auth/;
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
proxy_set_header Host $host;
proxy_redirect off;
}
location /docs/ {
proxy_pass http://docsapp/docs/;
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
proxy_set_header Host $host;
proxy_redirect off;
}
location /media/ {
add_header Access-Control-Allow-Origin *;
alias /usr/local/src/app/media/;
internal;
}
location /static/ {
alias /usr/local/src/app/static/;
}
client_max_body_size 8M;
}
Location \ is Vue.js app, remaining is Django REST Framework. I've searched a web for a while but none of found solutions worke for me.
Another problem occurred during attempt to split this file into two:
vue.conf
server {
listen 80;
location / {
alias /usr/share/nginx/html/;
}
}
and
drf.conf
upstream docsapp {
server app:8000;
}
server {
listen 80;
location /admin/ {
proxy_pass http://docsapp/admin/;
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
proxy_set_header Host $host;
proxy_redirect off;
}
location /api/ {
proxy_pass http://docsapp/api/;
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
proxy_set_header Host $host;
proxy_redirect off;
}
location /api-token-auth/ {
proxy_pass http://docsapp/api-token-auth/;
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
proxy_set_header Host $host;
proxy_redirect off;
}
location /docs/ {
proxy_pass http://docsapp/docs/;
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
proxy_set_header Host $host;
proxy_redirect off;
}
location /media/ {
add_header Access-Control-Allow-Origin *;
alias /usr/local/src/app/media/;
internal;
}
location /static/ {
alias /usr/local/src/app/static/;
}
client_max_body_size 8M;
}
It also didn't work. Any help would be appreciated.
I think since you do not change request URI when you proxy requests to the docsapp upstream, you can replace four location blocks where proxy_pass directive used with the following one:
location ~ ^/(?:admin|api|api-token-auth|docs)/ {
proxy_pass http://docsapp;
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
proxy_set_header Host $host;
proxy_redirect off;
}
You can't split your server block in two, only one of them will work acting as default server (see the documentation).

Setting expire headers for static content with nginx

I'm using Nginx in front of Glassfish (java application server) to serve my static content. I'm using following setting to set the expire headers
location /javax.faces.resource/images/ {
proxy_pass http://xx.xxx.xx:8080/javax.faces.resource/images/;
proxy_set_header Host $http_host;
proxy_set_header X-Real-IP $remote_addr;
expires 365d;
add_header Pragma public;
add_header Cache-Control "public";
}
I have also CSS and JS files inside /javax.faces.resource/ and want to set another expire header. How can I achieve this without breaking the setting for images?
Another question: I noticed that sometimes there is many Cache-Control entries in Response header. Is there a possibility to reset the properties instead of using add_header? Thanks
EDITED
server {
listen xx.x.yy.xxx:80;
server_name mydomain.com www.mydomain.com ;
rewrite ^/(.*) https://www.$server_name/$1 permanent;
}
server {
listen xx.x.yy.xxx:443; ## listen for ipv4
server_name www.mydomain.com;
ssl on;
ssl_certificate /etc/ssl/www.mydomain.com.2018.pem;
# Path to an SSL certificate;
ssl_certificate_key /etc/ssl/www.mydomain.com.2018.key;
# Path to the key for the SSL certificate;
client_max_body_size 20M;
access_log /var/log/nginx/mydomain.com.access.log upstreamlog;
error_log /var/log/nginx/mydomain.com.error.log;
rewrite_log on;
location / {
proxy_pass http://xx.x.yy.xxx:8080;
proxy_set_header Host $http_host;
proxy_set_header X-Forwarded-Host $http_host;
proxy_set_header X-Real-IP $remote_addr;
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
proxy_set_header HTTPS "on";
}
location /nginx_status {
stub_status on;
access_log off;
# allow 1.1.1.1;
# deny all;
}
location /images/ {
proxy_pass http://xx.x.yy.xxx:8080/images/;
proxy_set_header Host $http_host;
proxy_set_header X-Real-IP $remote_addr;
expires 365d;
add_header Pragma public;
add_header Cache-Control "public";
#add_header X-Cache-Status $upstream_cache_status;
#add_header Strict-Transport-Security max-age=15768000;
}
location /resources/desktop/images/ {
proxy_pass http://xx.x.yy.xxx:8080/resources/desktop/images/;
proxy_set_header Host $http_host;
proxy_set_header X-Real-IP $remote_addr;
expires 365d;
add_header Pragma public;
add_header Cache-Control "public";
#add_header X-Cache-Status $upstream_cache_status;
#add_header Strict-Transport-Security max-age=15768000;
}
location /javax.faces.resource/images/ {
proxy_pass http://xx.x.yy.xxx:8080/javax.faces.resource/images/;
proxy_set_header Host $http_host;
proxy_set_header X-Real-IP $remote_addr;
expires 365d;
add_header Pragma public;
add_header Cache-Control "public";
}
}

NGINX force SSL for all but health check file?

I have a Rails app with an NGINX reverse proxy behind an AWS ELB. I am terminating SSL on the ELB and I have NGINX configured to force any attempt at HTTP to rewrite as HTTPS. This setup works fine, but I'm also serving the site through ECS, and since the ELB health check is on HTTP port 80, when it gets the redirect and returns 301, the ELB health check fails and the instance is deregistered.
How do I set up NGINX to send all but the health check file through HTTPS?
Here is my server block from nginx.conf:
server {
listen 80;
server_name localhost;
root /var/www/html;
location ~ ^elbcheck\.html$ {
proxy_set_header X-Real-IP $remote_addr;
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
proxy_set_header Host $http_host;
proxy_redirect off;
proxy_pass http://rails_app;
break;
}
location / {
proxy_redirect off;
proxy_next_upstream error;
if ($http_x_forwarded_proto != "https") {
rewrite ^ https://$host$request_uri? permanent;
}
try_files $uri $uri/ #proxy;
}
location ~* \.(jpg|jpeg|svg|png|gif|ico|css|js|eot|woff|woff2|map)$ {
proxy_cache APP;
proxy_cache_valid 200 1d;
proxy_cache_valid 404 5m;
proxy_ignore_headers "Cache-Control";
expires 1d;
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_set_header X-Forwarded-Proto https;
add_header X-Cache-Status $upstream_cache_status;
proxy_pass http://rails_app;
}
location #proxy {
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_set_header X-Forwarded-Proto https;
proxy_pass http://rails_app;
}
}
I had the same problem, and found this answer somewhere on the internet (don't have the source anymore, it was a while ago)
server {
listen 80;
set $redirect_to_https 0;
if ($http_x_forwarded_proto != 'https') {
set $redirect_to_https 1;
}
if ($request_uri = '/status') {
set $redirect_to_https 0;
}
if ($redirect_to_https = 1) {
return 301 https://$host$request_uri;
}
...
}
Found a simple answer that worked great at this post. Here is what #ceejayoz suggested there:
server {
location /elb-status {
access_log off;
return 200;
}
}
Seems to be working--ECS hasn't terminated my service due to failed health checks anymore.

How to configure NGINX not to cache specific URL?

I have a NGINX server as front-end cache server and I'd like to disable cache on specific urls.
Here is the configuration on NGINX:
proxy_cache_path /tmp/nginx levels=1:2 keys_zone=my_zone:10m inactive=120m max_size=1000m;
proxy_cache_key "$scheme$request_method$host$request_uri$is_args$args";
server {
listen 10.0.0.45:80 default_server;
server_name proxy2.jjd;
include /etc/nginx/default.d/*.conf;
location / {
client_max_body_size 20m;
proxy_cache my_zone;
proxy_cache_bypass $http_cache_control;
proxy_no_cache $http_pragma $http_authorization $cookie_nocache $arg_nocache;
add_header X-Proxy-Cache-NGINX $upstream_cache_status;
add_header X-Real-IP $remote_addr;
add_header Cache-Control "public";
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
proxy_set_header X-Forwarded-Proto https;
proxy_set_header X-Forwarded-Port 443;
proxy_set_header Host $host;
proxy_pass http://127.0.0.1:8080;
proxy_read_timeout 90;
proxy_connect_timeout 90;
proxy_redirect off;
}
}
Add the following location to avoid an url:
location ^~ /your-url/ {
add_header X-Real-IP $remote_addr;
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
proxy_set_header X-Forwarded-Proto https;
proxy_set_header X-Forwarded-Port 443;
proxy_set_header Host $host;
proxy_pass http://127.0.0.1:8080;
proxy_read_timeout 90;
proxy_connect_timeout 90;
proxy_redirect off;
}
It just assigns this location to the proxy and doesn't enable caching for it.
As I get it, you just need a nested location with a single string proxy_cache off; inside to disable caching for nested URLs. Like this:
location / {
proxy_cache my_zone;
proxy_cache_bypass $http_cache_control;
// other stuff related to proxying or other processing
location /do/not/cache/this/url/ {
proxy_cache off;
}
}
you can just specify location do proxy_pass only for disable cache
location /will/not/cache {
proxy_pass http://127.0.0.1:8080;
..set_header ..
}

How do I fix this nginx configuration with almost-duplicate proxied locations?

I have the following nginx configuration for one of my virtual servers:
upstream app_example_https {
server 127.0.0.1:1340;
}
proxy_cache_path /Users/jaanus/dev/nginxcache levels=1:2 keys_zone=S3CACHE:10m;
proxy_cache_key "$scheme$request_method$host$request_uri";
server {
listen 0.0.0.0:1338;
server_name localhost;
ssl on;
ssl_certificate /Users/jaanus/dev/devHttpsCert.pem;
ssl_certificate_key /Users/jaanus/dev/devHttpsKey.pem;
location = / {
proxy_http_version 1.1;
proxy_set_header Connection "";
proxy_set_header Host 'something.s3-website-us-east-1.amazonaws.com';
proxy_set_header Authorization '';
proxy_hide_header x-amz-id-2;
proxy_hide_header x-amz-request-id;
proxy_hide_header Set-Cookie;
proxy_ignore_headers Set-Cookie;
proxy_cache S3CACHE;
proxy_cache_valid any 60m;
add_header X-Cached $upstream_cache_status;
proxy_pass http://something.s3-website-us-east-1.amazonaws.com/;
}
location /static/ {
proxy_http_version 1.1;
proxy_set_header Connection "";
proxy_set_header Host 'something.s3-website-us-east-1.amazonaws.com';
proxy_set_header Authorization '';
proxy_hide_header x-amz-id-2;
proxy_hide_header x-amz-request-id;
proxy_hide_header Set-Cookie;
proxy_ignore_headers Set-Cookie;
proxy_cache S3CACHE;
proxy_cache_valid any 60m;
add_header X-Cached $upstream_cache_status;
proxy_pass http://something.s3-website-us-east-1.amazonaws.com/static/;
}
location / {
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_set_header X-Forwarded-Proto $scheme;
proxy_pass http://app_example_https/;
proxy_redirect off;
}
}
What this does in English:
There’s an nginx frontend which serves requests either from a static Amazon S3 site, or an application server.
All requests to / (site root) and /static are reverse-proxied from Amazon S3. All other requests are reverse-proxied from the application server.
Now, the problem: there are two almost identical Location blocks for the S3. This was the only way how I could make this configuration work, where two specific folders (root and /static) are served from S3, and everything else goes to the application server.
Two almost-identical blocks look dumb and are not scalable. When I add such folders, I don’t want to keep duplicating the blocks.
How do I merge the two locations into one Location block, while keeping everything working the same way?
You could put repeating part into external file and include it.
amazon.inc
proxy_http_version 1.1;
proxy_set_header Connection "";
proxy_set_header Host 'something.s3-website-us-east-1.amazonaws.com';
proxy_set_header Authorization '';
proxy_hide_header x-amz-id-2;
proxy_hide_header x-amz-request-id;
proxy_hide_header Set-Cookie;
proxy_ignore_headers Set-Cookie;
proxy_cache S3CACHE;
proxy_cache_valid any 60m;
add_header X-Cached $upstream_cache_status;
proxy_pass http://something.s3-website-us-east-1.amazonaws.com;
your config
location = / {
include amazon.inc;
}
location /static/ {
include amazon.inc;
}
location / {
# proxy to you app
}
If you prefer to keep all in one file, you could use this trick:
error_page 470 = #amazon;
location = / {
return 470;
}
location /static/ {
return 470;
}
location #amazon {
# proxy to amazon
}
You could use regexp to merge several locations together, but I would not recommend to do that because it's hard to read and understand and is less efficient than simple prefix locations. But, just as an example:
# NOT RECOMMENDED
location ~ ^/($|static/) {
# proxy to amazon
}

Resources