New peak traffic and XF has started running really slow or not at all on DO LAMP stack

H

HJW

Active member
The CPU has been at most 15% and memory 20% and upgraded to add more of both with no change to the speed and restarted.

Is there anything else I should check to find why it's so slow / not loading?
 
Thanks for your reply. I believe it was Apache as was getting err_timed_out.

Change from 8gb 4vcpus to 16gb and 6 and it still wasn't working but there was loads of CPU and RAM left (80-90%).

1'500 concurrent users, 240k posts and quite a bit of uploaded images.

Thinking about changing to nginx - but all articles on DO are written in 2013. Then possibly separating mysql to a different droplet and moving the media to DO spaces. Or am I going in the wrong direction?
 
Switch to centminmod LEMP with mariadb and php7.2+

CentminMod.com LEMP Fully Optimized Nginx web stack for CentOS, AlmaLinux, Rocky Linux

Nginx installer for AlmaLinux, Rocky Linux and CentOS. Shell Menu based install and management for optimized LEMP stack - Nginx, PHP-FPM and MariaDB MySQL
centminmod.com centminmod.com
No need to use spaces. Use volumes to add space if needed.
You can use a tool like datadog to figure out whats going wrong.
What addons are you usng?
 
Thanks, just using tapatalk addon.

Is there an easy way to switch to centminmod LEMP? Other than creating a new droplet and moving all the data over?
 
@MattW @eva2000

Tapatalk could well be an issue. I would start a new thread on that topic to find out if anyone using tapatalk is experiencing performance issues.
 
  • Like
Reactions: HJW
Was the same for my site. You first get complaints, but if you lay the reasons out clearly most will understand. Tapatalk is setup to get your members onto their platform. Not the other way around. The longer you use it, the more trapped you get. So the sooner you ditch it the better.
 
Alfa1 said:
Was the same for my site. You first get complaints, but if you lay the reasons out clearly most will understand. Tapatalk is setup to get your members onto their platform. Not the other way around. The longer you use it, the more trapped you get. So the sooner you ditch it the better.
Click to expand...
Thanks, yeah it's on my roadmap to get rid of it. But having a site that stays online it the priority at the moment.

Is there someone doing services that can transfer my site to centminmod lemp?
 
Alfa1 said:
@MattW is offering this service.
Click to expand...
Yup Matt offers services related to Centmin Mod. I do for my private/large clients too but my prices would be much higher than Matt's.

If you're use to using DigitalOcean, keep an eye out for Centmin Mod LEMP's listing on DigitalOcean's 1-Click Marketplace in future https://community.centminmod.com/threads/digitalocean-marketplace-for-1-click-applications.16835/. I have signed up as a vendor and waiting for approval etc and doing private automation testing for building DigitalOcean images with Centmin Mod LEMP stack preinstalled. Once on DigitalOcean marketplace, you'll be able to do 1 click new droplet creation for a server with Centmin Mod LEMP stack preinstalled :D:cool:

There's a few additional road blocks for Centmin Mod marketplace listing, you can vote here to make sure DO folks implement my suggested fixes https://ideas.digitalocean.com/ideas/DO-I-2983 ;)
 
HJW said:
Is there someone doing services that can transfer my site to centminmod lemp?
Click to expand...

Huh??? Have you actually confirmed via logs and such what really happened? This is awfully silly move if you haven't confirmed what the problem was.
 
eva2000 said:
Yup Matt offers services related to Centmin Mod. I do for my private/large clients too but my prices would be much higher than Matt's.

If you're use to using DigitalOcean, keep an eye out for Centmin Mod LEMP's listing on DigitalOcean's 1-Click Marketplace in future https://community.centminmod.com/threads/digitalocean-marketplace-for-1-click-applications.16835/. I have signed up as a vendor and waiting for approval etc and doing private automation testing for building DigitalOcean images with Centmin Mod LEMP stack preinstalled. Once on DigitalOcean marketplace, you'll be able to do 1 click new droplet creation for a server with Centmin Mod LEMP stack preinstalled :D:cool:

There's a few additional road blocks for Centmin Mod marketplace listing, you can vote here to make sure DO folks implement my suggested fixes https://ideas.digitalocean.com/ideas/DO-I-2983;)
Click to expand...
Ah thanks that's good to know, will it be much difference from the current install scripts? Just tried doing that on a test box and it took a while but seemed to install all ok although wasn't just what image to use and selected Centos 7.5 x64.

WSWD said:
Huh??? Have you actually confirmed via logs and such what really happened? This is awfully silly move if you haven't confirmed what the problem was.
Click to expand...
Well I know my LAMP setup isn't great and seems like a Centmin build would be better, although yes it could be down to tapatalk
 
HJW said:
Ah thanks that's good to know, will it be much difference from the current install scripts? Just tried doing that on a test box and it took a while but seemed to install all ok although wasn't just what image to use and selected Centos 7.5 x64.
Click to expand...
Yeah Centmin Mod is a different beast (in a good way) and performance wise will be better than most LAMP stacks and even better than some LEMP stacks out there but my advice is you never have to jump in deep end and switch your live sites over. You can setup test hourly billed VPS and make a copy of your site to test privately while leaving your live site untouched. Get a feel for it on test VPS and then decide. Probably best advice is posted on my forums https://community.centminmod.com/threads/guide-to-learning-more-about-centmin-mod.10838/
 
HJW said:
Well I know my LAMP setup isn't great and seems like a Centmin build would be better, although yes it could be down to tapatalk
Click to expand...

Without knowing for sure you're just making changes for potentially no reason. This is the reason logs exist.
 
WSWD said:
Without knowing for sure you're just making changes for potentially no reason. This is the reason logs exist.
Click to expand...
You're totally right.

Just looking at the logs and I would think the 'server reached MaxRequestWorkers setting' could be the cause but it only comes up 3 times in the couple of hours that there were issues with the site

Code: 
[Tue Mar 12 20:44:38.910047 2019] [php7:error] [pid 24273] [client 88.12.240.196:41242] PHP Fatal error:  Uncaught XF\\Db\\Exception: No such file or directory in /var/www/html/src/XF/Db/Mysqli/Adapter.php:160\nStack trace:\n#0 /var/www/html/src/XF/Db/Mysqli/Adapter.php(28): XF\\Db\\Mysqli\\Adapter->makeConnection(Array)\n#1 /var/www/html/src/XF/Db/AbstractAdapter.php(60): XF\\Db\\Mysqli\\Adapter->getConnection()\n#2 /var/www/html/src/XF/Db/Mysqli/Adapter.php(113): XF\\Db\\AbstractAdapter->connect()\n#3 /var/www/html/src/XF/Db/AbstractAdapter.php(467): XF\\Db\\Mysqli\\Adapter->escapeString('options')\n#4 /var/www/html/src/XF/Db/AbstractAdapter.php(445): XF\\Db\\AbstractAdapter->quote('options')\n#5 /var/www/html/src/XF/DataRegistry.php(138): XF\\Db\\AbstractAdapter->quote(Array)\n#6 /var/www/html/src/XF/DataRegistry.php(83): XF\\DataRegistry->readFromDb(Array, Array)\n#7 /var/www/html/src/XF/DataRegistry.php(226): XF\\DataRegistry->get(Array)\n#8 /var/www/html/src/XF/App.php(1643): XF\\DataRegistry->offsetGet('options')\n#9 /var/www/html/src/XF/Container.php(28): XF\\App->XF\\{closure}(Object(XF\\Container))\n#10 /var/www/html/src/XF/App.php(2385) in /var/www/html/src/XF/Db/Mysqli/Adapter.php on line 160, referer:
[Tue Mar 12 20:44:39.419886 2019] [mpm_prefork:notice] [pid 1674] AH00169: caught SIGTERM, shutting down
[Tue Mar 12 20:44:50.227945 2019] [mpm_prefork:notice] [pid 1630] AH00163: Apache/2.4.18 (Ubuntu) OpenSSL/1.0.2g configured -- resuming normal operations
[Tue Mar 12 20:44:50.228264 2019] [core:notice] [pid 1630] AH00094: Command line: '/usr/sbin/apache2'
[Tue Mar 12 20:45:00.391407 2019] [mpm_prefork:error] [pid 1630] AH00161: server reached MaxRequestWorkers setting, consider raising the MaxRequestWorkers setting
[Tue Mar 12 20:53:04.642346 2019] [mpm_prefork:notice] [pid 1630] AH00169: caught SIGTERM, shutting down
[Tue Mar 12 20:55:52.805490 2019] [mpm_prefork:notice] [pid 1660] AH00163: Apache/2.4.18 (Ubuntu) OpenSSL/1.0.2g configured -- resuming normal operations
[Tue Mar 12 20:55:52.805779 2019] [core:notice] [pid 1660] AH00094: Command line: '/usr/sbin/apache2'
[Tue Mar 12 21:31:14.726483 2019] [mpm_prefork:notice] [pid 1660] AH00169: caught SIGTERM, shutting down
[Tue Mar 12 21:31:25.442162 2019] [mpm_prefork:notice] [pid 1692] AH00163: Apache/2.4.18 (Ubuntu) OpenSSL/1.0.2g configured -- resuming normal operations
[Tue Mar 12 21:31:25.442572 2019] [core:notice] [pid 1692] AH00094: Command line: '/usr/sbin/apache2'
 
Why have all that memory doing nothing? Optimize your InnoDB settings to use 80% of your RAM and use query caching.

I use DO with block storage for all images and attachments and don't have any issues but I don't have nearly that many concurrent users. I do use Litespeed as well, which is a very fast alternative to Apache, especially for php setups.
 
You must log in or register to reply here.
Back
Top Bottom