Has anyone tested this panel with more than 5000 users?

[ismaell14]

Hello, has anyone tested this panel with about 5000 peers connected at once?
I have been able to test it, everything was fine but in some moments I noticed instability and cut in the traffic of some peers, I have a quite powerful server, the bandwidth are at 30% of the total of this server, if someone could test it at this scale I would like to have your experience with this panel.

I appreciate your work for this panel, thank you

[eduardogsilva]

Hello @ismaell14,

I haven't tested the panel at this scale, but I don't foresee any inherent issues with handling such an installation.

Here are a few points to consider:

• Web Interface:
  The web interface runs on Django with SQLite. When not actively used, the system should remain mostly idle, consuming minimal CPU and only enough RAM to keep it running.

• Traffic Interruptions:
  When you mention "cut in the traffic," are you referring to missing data in the RRD graphs? This behavior is expected if the WireGuard service is restarted. The traffic counters reset to zero, and I've configured RRDtool to discard data when counters drop instead of when they rise, which helps reduce false traffic spikes on the graphs.

• Resource Usage:
  When the web interface isn’t in use, the only significant processes are the WireGuard service and RRDtool. It might be useful to monitor your resource usage for any CPU spikes, particularly during web interface interactions and the RRDtool cycles (which run every 5 minutes).

• Refresh Interval:
  Increasing the web refresh interval can reduce the load on both your web browser and the server.

For large-scale deployments like this, the web interface might be the most resource-intensive part—tools like Chart.js can sometimes struggle on less powerful devices. However, the WireGuard service itself should operate without major issues.

Also, keep an eye on your logs, resource usage and iops.

How is your browser dealing with this many peers?

I hope this helps, if you find anything that can be improved for an installation on this size, let me know.

Cheers!

[ismaell14]

what do you think would be an ideal value for web refresh interval?
thanks

[eduardogsilva]

I really don't know.
The default value is 10 seconds.

This means that every 10 seconds, the JS will query the /api/wireguard_status/ endpoint, and will update every peer statistics and realtime traffic graph for each peer.

this would be irrelevant on smaller environments, but honestly, I have no clue on how 5.000 chartjs graphs would work on the browser.

If this is lagging your pc, maybe you want to increase the value to 60seconds or more, on the other side, it will take longer to update the the peer list information.

[ismaell14]

I set this value to 100, I think that now it goes more smoothly when there are many connections,

Can you implement that when creating or deleting a peer to avoid the restart of the interfaces, that is done hot?

Thanks

[eduardogsilva]

I already took a look on something that could solve this issue.

I will try to implement a reload button as an alternative using wg syncconf. I believe that this will work.

[eduardogsilva]

I set this value to 100, I think that now it goes more smoothly when there are many connections,

Can you implement that when creating or deleting a peer to avoid the restart of the interfaces, that is done hot?

Thanks

Issue created here

[eduardogsilva]

Hello @ismaell14

I've just released an update that allows to reload instead of restarting.

Let me know how it goes for you.

Cheers!

[ismaell14]

hello, it works fine, but it would be perfect, to be able to choose which buttons appear to each user manager, to only let the peer manager the reload button, and thus prevent them from doing a restart. thanks for your great work.

[eduardogsilva]

That's a good suggestion. I will open an issue to check this later.

[eduardogsilva]

@ismaell14 Hello, just to let you know that this feature is now live in the latest version ;)

Cheers!

[ismaell14]

@ismaell14 Hello, just to let you know that this feature is now live in the latest version ;)

Cheers!

excellent tested and working perfectly, thanks for your work.

[ismaell14]

hi how are you, would it be possible to add a log section, where you could check if any peer was restarted, reloaded or if something happened with it?
I have some very punctual problems when I have many active peers running, this would be good to better identify that problem, thanks.

[eduardogsilva]

Hello, can you please start a new discussion explaining your use case and what you would like to have on the logs?