Plesk issues with trying to run UNA - APACHE/NGINX Issues (solved go to bottom for my current nginx directives)(Still Upstream Errors)

I seen media manager and the photo editor teases and I had to come back. I was getting a little discouraged and the media manager is a feature I have been wanting for a long time so happy to see it is in the works. I decided to not use Cyberpanel this time and use PLESK instead. Still having issues so site is inoperable. If you had similar issues with Plesk, please share and help if able.

**Update I got cron working, just did full path to the cron file and it is good. Noticing every time I log in and out in multiple browsers I have to refresh though. That is not normal. Same thing with going into studio sometimes have to relog in like 5 times. Logs showing no errors. 

I am going to paste what troubleshooting I have done so far just for help for other in case they have issues as well. 

Kyle

05:33

I even uninstalled just now/ reinstalled and did so with both PHP7.4 and php 8.0 and same exact thing. I wonder if it is the cron, but everything is pointing at it running correctly. I am absolutely stumped. Cron logs shows the cron job running every minute as it should be. Both when I had it on 7.4 and now back on 8.0. However I am still still the error up above I am realtime watching the log as I log out, refresh, login, refresh and that same error. Each install I have completely cleared the database. 

Everything at this point is surrounding this error I believe: [proxy_fcgi:error] [pid 84104:tid 139710521407232] [client IP :0] AH01071: Got error 'PHP message: PHP Fatal error: Uncaught Error: Call to a member function genDbCacheKey() on null in /var/www/vhosts/domain/httpdocs/inc/classes/BxDolLiveUpdates.php:200\nStack trace:\n#0. 

Here is the full log entry every time same error: 

[Sat Sep 25 08:42:05.998139 2021] [proxy_fcgi:error] [pid 19779:tid 140580906587904] [client IP0] AH01071: Got error 'PHP message: PHP Fatal error:  Uncaught Error: Call to a member function genDbCacheKey() on null in /var/www/vhosts/domain/httpdocs/inc/classes/BxDolLiveUpdates.php:200\nStack trace:\n#0 

/var/www/vhosts/domain/httpdocs/inc/classes/BxDolLiveUpdates.php(193): BxDolLiveUpdates->_getCacheKey()\n#1 

/var/www/vhosts/domain/httpdocs/inc/classes/BxDolLiveUpdates.php(205): BxDolLiveUpdates->_getCacheInfo()\n#2 

/var/www/vhosts/domain/httpdocs/inc/classes/BxDolLiveUpdates.php(233): BxDolLiveUpdates->_getCached()\n#3 

/var/www/vhosts/domain/httpdocs/inc/classes/BxDolLiveUpdates.php(113): BxDolLiveUpdates->_getCachedSystems()\n#4 

/var/www/vhosts/domain/httpdocs/live_updates.php(16): BxDolLiveUpdates->perform()\n#5 {main}\n  thrown in 

/var/www/vhosts/domain/httpdocs/inc/classes/BxDolLiveUpdates.php on line 200', referer: https://domain/

I tried and and had Leonid helping me too, but still same thing. It is fine until you do a login/logout action and then it is all refreshing from there. At this time it is basically the barebones, I have not installed anything no modules or anything at this point. Plesk does use Nginx and Plesk and does caching by default so this could be a big part of it. 

I had Plesk take a look and this is what they shared after doing the same troubleshooting we tried: 

Thank you for providing additional details, I was able to reproduce it.

Now allow me to summarize my findings.

In regards to the reported issue, the PHP-FPM pool(/var/log/plesk-php80-fpm/error.log) throws an exception(that is being redirected to the apache2 logs):

[26-Sep-2021 08:19:39] WARNING: [pool myvybz.com] child 45723 said into stderr: "PHP message: PHP Fatal error:  Uncaught Error: Call to a member function genDbCacheKey() on null in /var/www/vhosts/myvybz.com/httpdocs/inc/classes/BxDolLiveUpdates.php:200"
[26-Sep-2021 08:19:39] WARNING: [pool myvybz.com] child 45723 said into stderr: "Stack trace:"
[26-Sep-2021 08:19:39] WARNING: [pool myvybz.com] child 45723 said into stderr: "#0 /var/www/vhosts/myvybz.com/httpdocs/inc/classes/BxDolLiveUpdates.php(193): BxDolLiveUpdates->_getCacheKey()"
[26-Sep-2021 08:19:39] WARNING: [pool myvybz.com] child 45723 said into stderr: "#1 /var/www/vhosts/myvybz.com/httpdocs/inc/classes/BxDolLiveUpdates.php(205): BxDolLiveUpdates->_getCacheInfo()"
[26-Sep-2021 08:19:39] WARNING: [pool myvybz.com] child 45723 said into stderr: "#2 /var/www/vhosts/myvybz.com/httpdocs/inc/classes/BxDolLiveUpdates.php(233): BxDolLiveUpdates->_getCached()"
[26-Sep-2021 08:19:39] WARNING: [pool myvybz.com] child 45723 said into stderr: "#3 /var/www/vhosts/myvybz.com/httpdocs/inc/classes/BxDolLiveUpdates.php(113): BxDolLiveUpdates->_getCachedSystems()"
[26-Sep-2021 08:19:39] WARNING: [pool myvybz.com] child 45723 said into stderr: "#4 /var/www/vhosts/myvybz.com/httpdocs/live_updates.php(16): BxDolLiveUpdates->perform()"
[26-Sep-2021 08:19:39] WARNING: [pool myvybz.com] child 45723 said into stderr: "#5 {main}"
[26-Sep-2021 08:19:39] WARNING: [pool myvybz.com] child 45723 said into stderr: " thrown in /var/www/vhosts/myvybz.com/httpdocs/inc/classes/BxDolLiveUpdates.php on line 200"

Hence, the _getCacheKey($sType) returns a variable without any value(e.g. zero) to the listening function. Further questions in regards to this behavior are to be addressed to the application developers.

I would also have to add the following information to your request. Using a chrome tab with  option checked, the issue wasn't reproduced.

Hence, the application indeed supplies with an incorrect cache.

Now in regards to your other thoughts:

a) It is possible to change the myvybz.com from Apache2+Nginx to Nginx only hosting following the steps from the next article:

1. Log in to Plesk.

2. Go to Domains > domain > Apache & nginx Settings.

3. In the Nginx settings section, disable the Proxy mode setting.

b) By default Plesk is using brotli compression via Nginx, and I can see it enabled:

# curl -ILs https://domain -H "Accept-Encoding: gzip,br" | grep content-encoding
content-encoding: br

It can be disabled by adding an extra Nginx directive as described here:

1. Log in to Plesk.

2. Go to Domains > domain > Apache & nginx Settings.

3. Add an extra Nginx add directive: "brotli off;"

In addition. For some reason, the application is making attempts to reach the apache2ctl binary:

[26-Sep-2021 08:16:59] WARNING: [pool domain] child 45524 said into stderr: "which: no apachectl in ((null))"
[26-Sep-2021 08:16:59] WARNING: [pool domain] child 45524 said into stderr: "which: no apache2ctl in ((null))"
[26-Sep-2021 08:16:59] WARNING: [pool domain] child 45524 said into stderr: "which: no /usr/local/apache/bin/apachectl in ((null))"
[26-Sep-2021 08:16:59] WARNING: [pool domain] child 45524 said into stderr: "which: no /usr/local/apache/bin/apache2ctl in ((null))"

Meanwhile, this isn't connected to the reported issue, it is a good security question to the application developers.

Hence, summarizing the investigation results:

There is a code issue, that makes a child process crash when handling a cache. In case switching to Nginx only mode and disabling the brotli cache doesn't help here, please consider contacting the application developers in regards to the PHP error within the log files. 

Latest Update: If you disable Apache Proxy and go straight NGINX it will not load images or load your uploaded images, but there is no longer a cache issue. Just all images end up being not found even if you upload them. At this point I guess I need to go full NGINX and figure out how to setup a config, always have used Apache. Anyone have a sample config they can share? Tried the one here by Alex https://unacms.com/page/view-discussion?id=133, but images are not loading. 

Decided not to due to the full path and do the following: 

location / {

            index index.html index.htm index.php;

            rewrite "^/page/(.*)$" /page.php?i=$1 last;

            rewrite "^/m/(.*)$" /modules/index.php?r=$1 last;

            rewrite "^/s/([a-zA-Z0-9_]+)/([a-zA-Z0-9\.]+)" /storage.php?o=$1&f=$2 last;

            if (!-e $request_filename ) {

                rewrite  ^/(.+)$  /r.php?_q=$1  last;

                break;

            }

        }

If you use Plesk, just edit the following and do not worry like me, red it not bad if the code is correct: 

If anyone know how to get GZIp to work with NGINX etc please share if you know. 

What I got so far in my Additional nginx directives box (Was advised to make all changes here so make sure formatting is exact to not get errors) : As you can tell in Plesk it knew already being these settings are for this domain so no need to put path. 

location / {

    index index.html index.htm index.php;

    rewrite "^/page/(.*)$" /page.php?i=$1 last;

    rewrite "^/m/(.*)$" /modules/index.php?r=$1 last;

    rewrite "^/s/([a-zA-Z0-9_]+)/([a-zA-Z0-9\.]+)" /storage.php?o=$1&f=$2 last;

    if (!-e $request_filename ) {

        rewrite  ^/(.+)$  /r.php?_q=$1  last;

        break;

    }

}

# PROXY SETTINGS

proxy_buffer_size          128k;

proxy_buffers              4 256k;

proxy_busy_buffers_size    256k;

# FASTCGI

fastcgi_buffers 16 16k;

fastcgi_buffer_size 32k;

# GZIP COMPRESSION

gzip on;

gzip_proxied any;

gzip_types text/plain text/xml text/css application/x-javascript;

gzip_vary on;

gzip_disable “MSIE [1-6]\.(?!.*SV1)”;

Brotli is enabled by default. I wanted to give it a try so I re enabled it after disabling it earlier. This will be completely up to you. I will update if I run into any issues with it. Compared to GZIP it has better compression in terms of size and better performance. 

To keep GZIP do the following in SSH: sudo plesk bin nginx -d brotli to disable Brotli 

If going Brotli do not do anything it is already there in latest version of PLESK. If you accidentally disable it or change your mind do the following in SSH: sudo plesk bin nginx -e brotli

I noticed GZIP still loading so added the following directives to bottom: 

gzip off;

brotli on;

Due to some modules I use and love, I also reverted back to PHP 7.4 for now. 

I now ran into more issues with timeout in Apps Market and other areas of site so did following: 

#headers sesion time

proxy_connect_timeout 900;

proxy_send_timeout 900;

fastcgi_send_timeout 900;

fastcgi_read_timeout 900;

Still dealing with an upstream issue I can not figure out. It is still there when installing modules sadly not sure why no blocked ports it would not go through at all. Guess at this point I will never know, but for other who have PLESK/NGINX in the future here are my NGINX directives: 

location / {

    index index.html index.htm index.php;

    rewrite "^/page/(.*)$" /page.php?i=$1 last;

    rewrite "^/m/(.*)$" /modules/index.php?r=$1 last;

    rewrite "^/s/([a-zA-Z0-9_]+)/([a-zA-Z0-9\.]+)" /storage.php?o=$1&f=$2 last;

    if (!-e $request_filename ) {

        rewrite  ^/(.+)$  /r.php?_q=$1  last;

        break;

    }

}

# PROXY SETTINGS

proxy_buffering on;

proxy_buffer_size 128k;

proxy_buffers 4 256k;

proxy_busy_buffers_size 256k;

# FASTCGI

fastcgi_buffers 16 16k;

fastcgi_buffer_size 32k;

#headers sesion time

proxy_connect_timeout 900;

proxy_send_timeout 900;

fastcgi_send_timeout 900;

fastcgi_read_timeout 900;

gzip off;

brotli on;

Alex T⚜️ LeonidS 

#nginx #apacheproxy #plesk #fastcgi #upstream