Feb 16 10:40:24 nginx 2022/02/16 10:40:24 [error] 19557#100231: send() failed (54: Connection reset by peer)
-
Feb 16 10:40:24 nginx 2022/02/16 10:40:24 [error] 19557#100231: send() failed (54: Connection reset by peer)I am seeing this a lot as well as the sshguard entries that I have posted about yesterday. Using the 22.01 version. I did not remember seeing either of these entries on 2.5.2
-
nginx errors like that are pretty common if you have the dashboard open and then disconnect. I would not worry seeing them.
Steve
-
@troysjanda I have this problem too it makes my captive portal site loading slow.
Maybe the following is going on (see link below). I don't know how to fix it.
https://www.ateamsystems.com/tech-blog/nginx-service-nginx-upgrade-kill-sigusr2-results-shutdown-failed-54-connection-reset-peer/ -
Unlikely. Are you seeing nginx being restarted frequently?
-
@stephenw10 No I can't say i see nginx restarts but i see often things like on the picture below after i loaded the captive portal website. It takes a while for the page to show up. Before this page was blazing fast but not anymore. I must gave done something i don't know what to cause this. It seems like some sort of caching is not working anymore it takes several seconds to load a simple captive portal page.
-
Hmm, that error implies the client is not accepting the reply from nginx. It probably times out and retries introducing the delay.
Are you seeing it on all clients? All browsers?
-
@stephenw10 Those error's were created while using a wired desktop system on my LAN and using the latest stable Firefox browser. I am now on a laptop on a different vlan and wireless.
The error doesn't show up now but the loading of the captive portal is still slow.I didn't test any other browsers yet, wil do that tomorrow on the same desktop system.
I just tested with Edge and Vivaldi browser on the laptop, both load the captive portal slowee then normal and no errors in the general log thoug.
-
@stephenw10 I just realized i can boot the wired system remotely. So i just tested on the wired machine the captive portal page with Firefox, Edge and Vilvaldi browser.
The error didn't show up with all three the browsers. I think the slow loading captive portal page and those nginx errors are separate things.
It takes 6 seconds to load the captive portal page.
-
Could be an IPv6 issue. If the client tries to resolve/connect via IPv6 it will fail and have to timeout before trying IPv4. Does it have a routable IPv6 address?
-
@stephenw10 That could be very true.
My LAN is dual stack. The wlan1 were is captive portal is enabled is IPv4 only.I just tested this and you are 100% right.
Wen i use IPv4 the captive portal page is blazing fast again.Thank you again Stephen
-
@stephenw10
Does the captive portal work with IPv6 / dual stack? -
Not yet, there is no IPv6 support in the captive portal.
There are open feature requests for it and since the move from ipfw to pf for the layer 2 filtering that changes the situation. It would be in 23.05 at the earliest though.
Steve
-
