WebUI GET requests sporadically painfully slow
-
Hello,
I have built a forbidden router (pfsense operating in a VM, inside a proxmox host), I will take my lashings for such activity and continue on.
However, I'm having issues where sporadically GET requests to load a new page in the webUI take 10-30s to load, or fail entirely.
I have dug through the forums, and reddit and come with a few steps to remediate.
Originally the issue used to be every 2nd or 3rd page load would be slow, but now it's every 6-7.Current relevant configurations
Hardware config
CPU: 4 core [Host CPU Type] (Intel-11500T)
RAM: 4GiB
vDisk: 32G (Samsung 970evo NVMe)Software Config from other fixes
DNS
Primary 1.1.1.1, Secondary 9.9.9.9
Resolution behavior: use local, fallback to remote
(I use a pihole for everything else on my network)
DNS resolution in the gui worksFirewall logs widget is not on the dashboard, but this also happens when loading any page
Config file is only 108kbServices in use:
2 L2TP tunnels (to unifi UDM-Pros)
4 vLans, 1 Wan
OpenVPN setup and configured for mobile device access
FRR installed but not enabled
issue was happening prior to installing frr, and why I haven't configured/enabled it yet... too much loading on each setting page -
pfSense 2.7.2 ?
A VM so you have an easy console access. Or use SSH.
Option 8, and thentail -f /var/log/system.log /var/log/nginx.log
from now on all GUI web server pages requests are logged.
When the web GUI is slow, most often the DNS requests emitted by pfSense itself take 'ages' to get resolved.
grep 'start' /var/log/resolver.log
unbound isn't restarting to often ?
Or, as it is a VM after all : enough 'CPU' resources are allocated ?
-
@Gertjan Thank you for the reply!
I want to go back and test this again later but it seemed when I ran the tail log in your first command (in ssh) the gui was more responsive. I really had to zip (10-15 consecutive page loads) around the gui with the nginx/system log both visible for it to actually lock up.I also double checked the unbound log, and manually restarted it to make sure it was catching a start. Unfortunately, the only 'start of service' message I received was the manual restart from the status > services page.
I will add, while zipping through the highest CPU usage of the vm was 4.56%, I don't think that would be an evident issue at the moment but I'm open to suggestions.
Additionally, being that this is a webgui issue. I'm currently using firefox ver: 134.0b10 (64-bit), I switched to beta during the youtube issue.
This does appear on chrome as well.
Outside of browser, I previously had the VM configured with full passover of the realtek (RTL8125) nic for the WAN interface, and a virtIO linux bridge for the LAN adapter. I stood up/rebuilt the pfsense vm, and re-added the config from a backup recently in an attempt to fix this issue, and both adapters are currenly virtIO linux bridge adapters.