friendica (DFRN) - Link to source

Delay loading /network

!Friendica Support
Hi everyone, I have a big problem on my Friendica installation , When I connect to niwego.com/network , it takes 30/40 seconds to load the page,I chose to display only 10 post on the page, plus when I scroll it lags a lot.

I installed the plugin rendertime , this is the result , it has become almost unusable

Database: 3.619/0, Network: 0, Rendering: 40.65, Session: 0, I/O: 0.01, Other: 0.27, Total: 44.54
Class-Init: 0.005, Boot: 0.004, Init: 0, Content: 44.524, Other: 0.017, Total: 44.55

Ubuntu Server 24.04 LTS

This is the third installation of Friendica, on the others I have the delay, but on this installation (my personal one) it is too much.
Friendica runs on a virtual machine with 8 Cores and 8Gb of memory

Some suggestions , thanks...

!Friendica Support
in reply to Anomaly

friendica - Link to source

Scifijunkie

Well, it's back to not loading /network or /community/global and that's with CloudFlare set to DNS only when it comes to Friendica. The odd thing is that when I try to go to /network or /community/global it takes out that entire node of my k3s cluster and it has three nodes. I could see if there is an update to k3s and see if that does anything.
in reply to Anomaly

friendica - Link to source

Scifijunkie

Well, I did a rather crazy thing. I looked to see if I could find something like htop but for hard drives. I found something called iotop. Now after running it I noted that the total and current disk read was going all over the place. I went ahead and tried to load /network while iotop was running and it would spike to 20 M/s. So I ran e4defrag and let it sit for a while. It finely settled to about 200 K/s. I went ahead and tried to load /network again. It spiked to about 2 to 4 M/s. I saw that some people reported that PeerTube was doing a slow down to so I loaded up my PeerTube which defaults to me to videos just on my PeerTube node. It spiked to 40 M/s.

Now the only real difference between the two to me is that I use lscr.io/linuxserver/mariadb:10.11.5-r0-ls123 when it comes friendica:stable and I use postgres:15 when it come to chocobozzz/peertube:production-bookworm. So it makes me wounder if I need to update mariadb.

in reply to Anomaly

friendica - Link to source

Scifijunkie

Well, I updated mariadb from lscr.io/linuxserver/mariadb:10.11.5-r0-ls123 to lscr.io/linuxserver/mariadb:10.11.8-r0-ls153. It still spikes but between about 2.5 to 4 M/s and it still takes out the k3s node it's on after a while. At least it seems to still take out the k3s node it's on anyway. That is when I try to load /network.
This entry was edited (Thursday, August 29, 2024, 3:26 AM)
in reply to Anomaly

friendica - Link to source

Scifijunkie

in reply to Anomaly

friendica - Link to source

Scifijunkie

Well it seems that @Roland Häder may have been right about that I may have inadvertently caused my own problem. I switched my Interface setting in my Pihole from "Bind only to interface enp3s0" to "Allow only local requests" thinking it would address the "Maximum number of concurrent DNS queries reached" warning better than just upping the concurrent DNS queries. Well it worked by making the warning but I think it might have created the problem of the /network or /community/global not loading for some resone. I say that I switched my Pihole back to "Bind only to interface enp3s0" and I'm having no problems loading either /network or /community/global. I have not looked to see if the spikes still happen or not though when it comes to the mariadb Database it uses. Just a heads up for other admins that use Pihole for DNS inside their networks.
@Roland Häder🇩🇪
This entry was edited (Sunday, September 1, 2024, 5:59 AM)
⇧