App unresponsive due to error

[privsec]

It may be due to a hacking attempt

https://support.plesk.com/hc/en-us/articles/115002643313-WordPress-site-hosted-in-Plesk-is-slow-Lots-of-log-entries-POST-xmlrpc-php-HTTP-1-0-499

[imc67]

@privsec thanks for finding however the xmlrpc on my sites are always locked for security reasons.

However, after searching around the internet and this forum I combined some answers and tried the following:

In the filemanager there is a folder /apache/ and it contains a file mpm_prefork.conf.

There I changed the (very low values) to:

StartServers	10

MinSpareServers	10

MaxSpareServers 40

MaxRequestWorkers	  256

MaxConnectionsPerChild   1000

Let’s wait and see if this solves the issue.

[privsec]

@imc67 how much daily traffic do you get?

[nebulon]

@imc67 do you have any plugin installed which may use websockets? Those are long-lived, so each websocket connection will use one request worker.

[girish]

@imc67 said in App unresponsive due to error:

mpm_prefork.conf

Indeed, fixing the values in that file is the correct fix.

[imc67]

@nebulon said in App unresponsive due to error:

do you have any plugin installed which may use websockets

I checked the inspector but couldn't find any

[imc67]

@girish said in App unresponsive due to error:

Indeed, fixing the values in that file is the correct fix.

The changed values result finally in a more stable situation.

This is not a high traffic site (5k month) and the only one created with Avada Theme and that's really a "monster".

[d19dotca]

I didn't notice any unresponsiveness but combing through WP logs for a different issue for a Shop Manager user, I noticed a few instances of the same error reported in this thread. It's a large WooCommerce site being built and they're making a lot of product changes so this is what I suspect the root cause of that is in my case at least.

So it's safe to increase that then eh? Any guidance on increasing that in terms of how it may affect Cloudron at all?

@imc67 - what did you increase yours to out of curiosity?

[imc67]

@d19dotca said in App unresponsive due to error:

@imc67 - what did you increase yours to out of curiosity?

This my current config, it's 99% more stable than before:

<IfModule mpm_prefork_module>
    # On startup, start these many servers
	StartServers	10

    # At any given time, keep atleast these many servers
	MinSpareServers	10

    # At any given time, keep atmost these many idle servers (this is always >= MinSpareServers+1)
	MaxSpareServers 40

    # Maximum number of servers at any given instant. Requests will be queued after this
	MaxRequestWorkers	  256

    # Recycle process after handling these many requests. This protected against accidental memory leaks
	MaxConnectionsPerChild   1000
</IfModule>

[d19dotca]

Perfect! I ran some tests today with tweaking various numbers and looked at recommendations online too. It's mostly a copy of yours @imc67 in the end but with a lower MaxConnectionsPerChild. Ran a bunch of different tests with different numbers but concluded that was the best bang for the buck so-to-speak given the 4 GB memory max set on each WP app.

It's a mix of default values from the Ubuntu install and the default Apache values in the original package too (for the 256 MaxRequestWorkers for example). I'm pleased with the results.

Here's what I have for reference now, and I used ab (Apache Benchmark) to run some performance tests and had significant improvements over the defaults. Thought I'd share the testing results in case anyone was interested.

New mpm_prefork.conf I'm using:

# Restart the app if you make changes to this file

<IfModule mpm_prefork_module>
    # On startup, start these many servers
	StartServers 5

    # At any given time, keep atleast these many servers
	MinSpareServers	5

    # At any given time, keep atmost these many idle servers (this is always >= MinSpareServers+1)
	MaxSpareServers 10

    # Maximum number of servers at any given instant. Requests will be queued after this
	MaxRequestWorkers 256

    # Recycle process after handling these many requests. This protected against accidental memory leaks
	MaxConnectionsPerChild 500
</IfModule>

Resources:

• https://httpd.apache.org/docs/2.4/mod/prefork.html
• https://exampleconfig.com/view/apache-ubuntu20-04-etc-apache2-mods-available-mpm_prefork-conf
• https://www.woktron.com/secure/knowledgebase/133/How-to-optimize-Apache-performance.html
• https://www.maketecheasier.com/optimizing-apache-part-1/

Results from ab with default settings that come with Cloudron's WP package: (note the rather large spread between min and max request times)

Concurrency Level:      50
Time taken for tests:   12.391 seconds
Complete requests:      1000
Failed requests:        0
Total transferred:      84813000 bytes
HTML transferred:       84205000 bytes
Requests per second:    80.70 [#/sec] (mean)
Time per request:       619.574 [ms] (mean)
Time per request:       12.391 [ms] (mean, across all concurrent requests)
Transfer rate:          6684.04 [Kbytes/sec] received

Connection Times (ms)
              min  mean[+/-sd] median   max
Connect:      227  291  36.4    287     401
Processing:   220  281 122.2    252    1039
Waiting:      144  190 120.0    161     954
Total:        451  572 127.3    546    1285

Percentage of the requests served within a certain time (ms)
  50%    546
  66%    570
  75%    584
  80%    594
  90%    624
  95%    818
  98%   1091
  99%   1261
 100%   1285 (longest request)

Results after tweaking to the above new configuration: (note that the max request time here is about 60% faster still than original config results and overall a much more consistent behaviour too!)

Concurrency Level:      50
Time taken for tests:   11.358 seconds
Complete requests:      1000
Failed requests:        0
Total transferred:      84813000 bytes
HTML transferred:       84205000 bytes
Requests per second:    88.04 [#/sec] (mean)
Time per request:       567.899 [ms] (mean)
Time per request:       11.358 [ms] (mean, across all concurrent requests)
Transfer rate:          7292.24 [Kbytes/sec] received

Connection Times (ms)
              min  mean[+/-sd] median   max
Connect:      230  287  26.9    283     389
Processing:   220  252  17.3    249     329
Waiting:      144  160  10.0    157     217
Total:        467  539  33.0    538     649

Percentage of the requests served within a certain time (ms)
  50%    538
  66%    550
  75%    559
  80%    567
  90%    584
  95%    599
  98%    618
  99%    622
 100%    649 (longest request)