I've tried to deploy two nodeJS apps on Digitalocean using a dokku droplet. I am using the "virtualhost naming" scheme but there is a problem.
My DNS configuration looks like this:
I have the main app and the admin app. I would expect to view the admin app when i visit app.example.com (I actually have a proper domain name) but I can see the same app when hitting example.com and app.example.com.
There is something wrong with nginx probably, but I don't know exactly what is going bad?
One thing I have noticed is that whichever app is installed first will be the one that example.com forwards to.
You are correct to attribute this behaviour to Nginx. I think it's due to it falling back to this config somehow when it doesn't detect a config for example.com
This dokku plugin (https://github.com/progrium/dokku/tree/master/plugins/nginx-vhosts) is responsible for rewriting the nginx.conf for each app every time it is deployed.
Nowadays it uses a template nginx.conf (https://github.com/progrium/dokku/blob/master/plugins/nginx-vhosts/templates/nginx.conf) although this is a fairly recent change so be sure your on a recent version.
You will end up with a Nginx config that looks like the following:
server {
listen [::]:80;
listen 80;
server_name app.example.com;
return 301 https://$host$request_uri;
}
I'm not currently sure why the above snippet results in the described behaviour. A work around is to setup your own nginx conf in /etc/nginx/sites-enabled/ with
server_name example.com;
but pointing to a holding page or whatever works for you.
Related
I'm trying to add a new virtualhost to host a separate website on my Gitlab on-prem version.
I have followed the steps in https://stackoverflow.com/a/39695791.
I'm trying to set up a reverse-proxy, added the custom_nginx_config direction got /etc/gitlab/gitlab.rb. it looks like the configuration is loaded because when I had a syntax error I could see the syntax issues via gitlabctl-tail.
I've fixed the syntax issue and now when I'm reconfiguring/restarting nginx and gitlab-ctl it does not seem to accept the new config.
the new configuration is as follows:
server {
listen *:80;
server_name domain.com;
location / {
proxy_pass http://localhost:4444
}
}
and then when I try to access domain.com I'm getting redirected to the gitlab app and not to my internal app
I have a published site that uses HTTPS. The site needs to communicates with a HTTP node Express API. The API is run on my local machine. Everything worked fine until I switched the client application to use HTTPS. Now I receive mixed content warnings. I have been reading about reverse proxys and wonder if this could be the solution to my problem. Is it possible to proxy a request to my localhost? Or will localhost point to the server the proxy is on?
I have been looking at using nginx as the reverse proxy server but I have zero experience with proxys and not positive how to go about it.
I am mainly wondering if it is possible or not before I dig any deeper.
Yes, this is a pretty standard use case for using nginx (or any other reverse proxy). You would configure the location prefixes, etc that need to go to your backend application and proxy (via proxy_pass directive) to them. Any static content can be served directly from nginx. All of this can then behind nginx.
Assuming that your application is never issuing absolute urls which make use of "http://" this should resolve your mixed content warnings.
You will probably want to read some tutorials but the basics of your configuration would be:
server {
listen 443 ssl; # you can also add http2
server_name hostnames that you listen for;
ssl_certificate_key /path/to/cert.key;
ssl_certificate /path/to/cert.pem;
root /var/www/sites/foo.com;
location /path/handled/by/application {
proxy_pass http://localhost:8000; # or whatever port is
}
}
I have 5 domains pointing to the same machine, let's say these domains are:
example1.com
example2.com
example3.com
example4.com
example5.com
My server has a virtual host file managed by apache, which controls these 5 domains.
I have a node.js app running on port 3000.
If I connect to any of the 5 domains, and type :3000 at the end, I can see the node.js app splash screen.
For example, if I connect to example1.com:3000 it serves me the same page (the node.js app) that can be reached by connecting to example2.com:3000 and example3.com:3000, etc. I understand why this is happening.
However, I am trying to edit my configuration such that the following criteria are met:
Users can only connect to the node app when they type in example5.com, and not any of the other domains.
Users do not have to type in the port number, example5.com:3000, when trying to reach the node app.
I have searched and found some information which leads me to believe this is normally achieved by setting up a reverse proxy, but most examples I find are using nginx or node. It would be nice to figure out a reverse proxy solution with apache so I don't have to re-write all my virtual host logic with node or nginx.
You need to set up a reverse proxy. An easy way to do this is to use nginx. Install and start nginx and in your config file put the following:
server {
listen 80;
server_name example1.com;
location / {
proxy_pass http://localhost:3000;
}
}
And then run your node server on port 3000.
I bought some domains at godaddy.com (i.e mydomain.com) for my droplet at digitalocean.com (i.e 199.216.110.210). I run a nodejs application on port 80 on the droplet. From godaddy.com, I forward with masking mydomain.com to 199.216.110.210 and I could see may app.
Now I want to run on 199.216.110.210 several node applications on different ports, using ngnix as reverse proxy. I followed the instructions here (www.digitalocean.com/community/articles/how-to-host-multiple-node-js-applications-on-a-single-vps-with-nginx-forever-and-crontab).
My nginx .conf file is
server {
listen 80;
server_name mydomain.com;
location / {
proxy_pass http://localhost:3000;
# same as in the link above
}
}
(and I am sure it is read: when ngnix start if I put an error there, ngnix reports it).
I start the nodejs application on port 3000:
I try mydomain.com, but ngnix shows always the welcome page.
Also doing mydomain.com: 3000 does not work,
it works only with 199.216.110.210:3000.
From godaddy.com, if I forward with masking the mydomain.com to 199.216.110.210:3000 I can see may app.
But I do not like this solution. I would like domains pointing to my droplet, without specifing the port and admin them with nginx.
How can I get a domain name to use with ngnix as reverse proxy to select my apps, mapped on different domains on different ports? I suppose that forwarding from godaddy.com is somehow limited.
In your server go to /var/log/nginx and do a tail -F *log. Now in another shell restart nginx.
I suspect that your domain name is too long and nginx will complain about its hash_bucket_size is too small. If this is the case open /etc/nginx/nginx.conf and make sure that the line
server_names_hash_bucket_size 64;
exists, has a value of 64 and is uncommented. Then do sudo service nginx reload, and check if all works as expected.
I am going to detail step by step how I am able to do it in my aws ec2 instance;
I set up a DNS record to my instance, so i can set mydomain.com to 192.168.123.123 (my specific IP).
Inside my instance I have forever running my node.js app in port 3000 (I test it work by issuing curl localhost:3000 from the command line)
I then download this .sh file in order to properly intantiate nginx; curl -o nginxStarter.sh https://gist.githubusercontent.com/renatoargh/dda1fbc854f7957ec7b3/raw/c0bc1a1ec76e50cdb4336182c53a0b222edb6c0e/start.sh
I configure nginx with this configuration file. Put this file in; /etc/nginx/nginx.conf
Start nginx with this command; sudo sh nginxStarter.sh start
PS.: For multiple apps just replicate the lines that routes the requests to specific ports, very easy...! If not needed you can eliminate lines regarding out SSL.
I need the right configuration of nginx for my problem.
Suppose the nginx + nodejs serverprograms are running on the same debian machine.
Domain name for my website is for simplicity just webserver.com (and www.webserver.com as alias)
Now, when someone surfs on the internet to "webserver.com/" it should pass the request to the nodejs application which should run on a specific port like 3000 for example. But the images and css files should get served by nginx as static files and the filestructure should looke like webserver.com/images or webserver.com/css .. images + css should get served by nginx like a static server
Now it gets tricky:
But when someone surfs on webserver.com/staticsite001 or webserver.com/staticsite002 then it should get served by the nginx server only. no need for nodejs then.
And for the nodejs server, I am just setting up my nodejs application with port 3000 for example to receive the bypass from nginx for webserver.com/
to put it in a more understandable language: when someone surfs to webserver.com/staticsite001 it should NOT pass it to the node application. It should only pass it to the node application if its inside of the first webserver.com/ directory that the outsiders can see. The webserver.com/staticsite001 should only get serverd by nginx.
How, how do I do that ? And what should the http and server block look like for the nginx configuration look like?
I am familiar with nodejs. But I am new to nginx and new to reverse proxying.
thanks
the file structure on the debian hard drive looks like:
/home/wwwexample/staticsite001 (for www.webserver.com/staticsite001/) only handled by nginx
/home/wwwexample/staticsite002 (for www.webserver.com/staticiste002/) only handlex by nginx
/home/wwwexample/images
/home/wwwexample/css
and in
/home/nodeapplication is my node js application
This server block should work:
server {
listen 80;
server_name webserver.com www.webserver.com;
root /home/wwwexample;
location / {
proxy_pass http://localhost:3000;
proxy_set_header Host $host;
}
location /staticsite001 {
}
location /staticsite002 {
}
location /images {
}
location /css {
}
}
First location makes nginx to proxy everything to localhost:3000. Following empty locations instruct nginx to use default behavior, that is to serve static files.
Put this code into file /etc/nginx/sites-available/my-server and create a symlink to it in /etc/nginx/sites-enabled. There is a default config, which you could use as a reference.
After that you could use command sudo /usr/sbin/nginx -t to check configuration. If everything is OK use /etc/init.d/nginx reload to apply new configuration.