At one of my hosting servers, I've a website that has a strange performance problem. Sometimes the server needs 8.2s to load the webpage, other times it's 200ms. It's almost never different/in between.
I've set up a monitoring script (Zabbix web) at 5 "comparable" websites on the same server and ran it for a few days. The graph shows me that there's only one virtual host that has this problem. Also a empty php script (or just a static file) experiences the same problem, so based on that I guess it isn't php related (even though I tested with different php versions but they all have this problem). Other vhosts don't have that same problem, they're always between a few ms and a few hundred, but that's "normal".
I compared the Apache-config: there's no difference between this and other websites.
I rebooted the server, restarted Apache, but the problem never disappears. I don't know where to look for anymore because based on my findings I guess it should be some specific config for this website, but I can't think of a place to look for.
Ubuntu 16.04 / Apache 2.4.18
This website doesn't have that many visitors (just a regular amount) and because of my Zabbix the website has two (single file) visits every minute.
Comments
Submitted by tfhelp on Thu, 08/08/2019 - 02:14 Comment #1
Submitted by Jfro on Thu, 08/08/2019 - 02:27 Comment #2
The .tld different.
NS dns lookups?
A extra service as cloudflare before and so on only thinking loud.
Submitted by tfhelp on Thu, 08/08/2019 - 02:49 Comment #3
Submitted by tfhelp on Thu, 08/08/2019 - 02:50 Comment #4
Submitted by tfhelp on Thu, 08/08/2019 - 03:29 Comment #5
Jfro: I just updated the issue when I tried to add more relevant information and I saw an dns-related issue (two A-records directing to different ip-adresses for the same domain name). You suggested to look at the same, so I guess we must be right, I'm waiting for my customer to solve this (of course, I can check using /etc/hosts but to be sure it's better to wait for the real dns change)
Submitted by Jfro on Thu, 08/08/2019 - 03:17 Comment #6
Ok thanks for update. Please post solution and cause here also if found? ( handig voor anderen die zoeken)
Submitted by tfhelp on Fri, 08/09/2019 - 07:25 Comment #7
Cause was indeed a double A-record in the DNS. This caused a kind of load balancing between our server and another (external) server. The external server didn't respond (I guess the IP didn't exist or something like that) en that caused a 8 second 'timeout'.
So, in case some-one experiences a 8 second timeout, please check your DNS settings.