Testing subfolder installation

[finid]

Continuing from this topic, I tested subfolder installation on a server running in a virtual environment.

The site mostly works, but I'm getting an error when first loading the front page. It will not load completely and just show a blank page. Eventually after much prodding, it will load.

On Firefox, the console error says:

 Firefox can't establish a connection to the server at ws://192.168.1.10/forum/socket.io/1/websocket/iGeXqL_AHEkKWOaKkQRF.

And on Chromium, the error shown in the console:

 WebSocket connection to 'ws://192.168.1.10/forum/socket.io/1/websocket/n4YHWE2OK7OPguiNkQQ-' failed: Error during WebSocket handshake: 'Connection' header value is not 'Upgrade' http://192.168.1.10/forum/

When it fails to load the front page, the console error is:

Failed to load resource: the server responded with a status of 504 (Gateway Time-out) http://192.168.1.10/forum/socket.io/1/xhr-polling/Ntkjy5eI8TikAr5akQRQ?t=1394536265747

Here's the part of the Nginx config relevant to this topic:

location ^~ /forum {

        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 ^/forum/(.*)$ /$1 break;

        proxy_pass http://192.168.1.10:4567;
        proxy_redirect off;

        # Socket.IO Support
        proxy_http_version 1.1;
        proxy_set_header Upgrade $http_upgrade;
        proxy_set_header Connection "upgrade";

    }

@julian, @psychobunny and @baris, what do you make of this?

Edit: Note that even though the front page does not load, clicking on the menu links work.

[finid]

Still more on this:

On the ACP, when I click on the Events button, the page fails to load. Instead, I get the notification shown in this screenshot.

But when I look under /var/www/forum, which is the root of the NodeBB installation, the events.log does not exist.

I think this events.log-related issue only exists in the Edge code, because I do not see it on my site running 0.3.2.

[julian]

What is relative_path in your config.json?

[finid]

@julian

/forum and that is the only reason why it worked. Without it, trying to access the site at http://192.168.1.10/forum always redirected to http://192.168.1.10/404.

[finid]

And on Firefox, the home and category pages will load to about 80%, and stop. Only after clearing the cache will the loading complete.

[finid]

On the ACP, the Events tab now loads, but there's still no log file in the root of the installation or under the logs directory.

[testerool]

i got this issue too.

[finid]

@testerool

Which one, missing events.log or the issue raised in the original post?

[testerool]

@planner missin events.log

[julian]

@planner That's right, with no events logged, it will just show that message. This page doesn't show errors.

[finid]

@julian

Does that also mean that the events.log file is automatically created only when events are logged? I expect that the file will at least be created and remain empty until events are logged.

[julian]

@planner The file is created when something is logged to it -- otherwise it is not created.

In unrelated news, latest commits have subfolder installs working again.

Notes

• My proxy config's "location" is actually /, not /forum/
• Git hash: 313c51cfb1fc1df2d44e5a2b5420fe87dd43968c
• There are still some oddities, but it should be 90% there.

[finid]

@julian

Yes, it's definitely 90% there, which is why I'm very happy with this.

However, if my relative path is / and not /forum, ti just does not work.

Btw, does "failed: Error during WebSocket handshake" reveal anything major?

[julian]

I used a relative_path of /forum as intended, but in my nginx config, the line reads location / {, not location ^~ /forum {

Perhaps that might get you closer?

Also, this error: Failed to load resource: the server responded with a status of 504 (Gateway Time-out) http://192.168.1.10/forum/socket.io/1/xhr-polling/Ntkjy5eI8TikAr5akQRQ?t=1394536265747 suggests that your nginx server is < 1.3.13... maybe?

[finid]

suggests that your nginx server is < 1.3.13... maybe?

@julian

Yes, Nginx is 1.2.6, but I'm not too worried about that because the production server is running the latest and greatest.

In my Nginx config I already have a location / { that does something else, so another one of those will throw a duplicate error. FWIW, location ^~ /forum { works for me. The only real issues are specific to NodeBB. Forever, in the ACP's Settings page, clicking on the tabs does not always load the correct tab.

[julian]

For the life of me, @planner, still am not seeing the problems you're encountering!

However, I did notice that I had to change the proxy_pass line in my config:

proxy_pass http://127.0.0.1:4567/forum/;

This, in addition to location ^~ /forum/ {, and ensuring that relative_path is set to /forum, and now my NodeBB loads properly:

[finid]

@julian

Well, mine loads properly too, even with proxy_pass setting:

proxy_pass http://127.0.0.1:4567;

And adding /forum at the end seems to make no difference.

The real issue I'm seeing is this: On the Settings page in the ACP, I'm not able to navigate between the tabs. No matter which one I click on, the General tab is still in focus. Can you verify this?

To illustrate, try matching the URL in this page with the page in focus.

[julian]

No, actually, I haven't been able to reproduce that specific bug

[finid]

@julian

Well, that's good news. I'll just attribute it to running the wrong version of Nginx (1.2.6) and assume that when I configure the production server to run from a subfolder, that all will be fine. but that won't happen until 0.4.0 is released.

[finid]

@julian,

I've been posting to a site that's supposed to be the production site once 0.4.0 is released. So I've been running it without the aid of Nginx, just off the built-in server that comes with NodeBB (I think that's Express) and accessing the site from http://linuxbsdos.com:4567

After testing the subfolder installation in a local installation, I decided to upgrade the soon-to-be production site to the latest code and configure Nginx to proxy serve it from a subfolder. Unfortunately, it's not working like the local installation. I'm getting a 520 Bad Gateway error.

My Nginx config and json.config are really no different from the ones I used in the local installation. But to make it easier for others to help me troubleshoot, here are the complete settings:

Nginx config. Note: I've tried proxy_pass http://linuxbsdos.com:4567/forum, but it still didn't work.

##
server {

    server_name linuxbsdos.com www.linuxbsdos.com;
    
    ## Your only path reference.
    root /public_html/;
    listen 8080;

    index index.html index.htm index.php;

    include conf.d/drop;

        location / {
                # This is cool because no php is touched for static content
                        try_files $uri $uri/ /index.php?q=$uri&$args;
        }

        location ~ \.php$ {
            fastcgi_buffers 8 256k;
            fastcgi_buffer_size 128k;
            fastcgi_intercept_errors on;
            include fastcgi_params;
            fastcgi_param SCRIPT_FILENAME $document_root$fastcgi_script_name;
            fastcgi_pass unix:/var/run/php5-fpm.sock;


        }

        location ~* \.(css|js|png|jpg|jpeg|gif|ico)$ {
                expires 1d;
        }

## Settings for proxying NodeBB forum installation. Added March 14 2014.

        location ^~ /forum {

        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://127.0.0.1:4567;
        proxy_redirect off;

        # Socket.IO Support
        proxy_http_version 1.1;
        proxy_set_header Upgrade $http_upgrade;
        proxy_set_header Connection "upgrade";

    }

}

And config.json

{
    "base_url": "http://127.0.0.1/forum",
    "port": "4567",
    "use_port": true,
    "secret": "secret",
    "bind_address": "0.0.0.0",
    "database": "redis",
    "redis": {
        "host": "127.0.0.1",
        "port": "6379",
        "password": "secret1",
        "database": "0"
    },
    "bcrypt_rounds": 12,
    "upload_path": "/public/uploads",
    "relative_path": ""
}

From the error logs, I'm getting entries like:

2014/03/14 22:13:15 [error] 9689#0: *339 connect() failed (111: Connection refused) while connecting to upstream, client: 127.0.0.1, server: linuxbsdos.com, request: "GET /forum/forum-43.html HTTP/1.1", upstream: "http://172.143.217.163:4567/forum/forum-43.html", host: "linuxbsdos.com"

I don't know where to look now, but I'm thinking that all the entries in the database have different URLs than what I'm trying to connect with. Is that a factor or should I be looking at other things?

Should the client IP shown in the error log be 127.0.0.1?

Hey, @julian, I've updated the settings a lit, basically, tried a little bit of everything, but this is not working in production. Any ideas, clues?

Btw, I'm running Nginx 1.4.6.