Posted: Wed Aug 27, 2025 17:32 Post subject: Slowdown / page not found after 6 to 8 hours after reboot
Hello everyone,
I have a setup with DNScrypt. Default values, nothing fancy. My router (Asus RT-AC68U on 61238) is hooked up to a 250MBit cable connection. After about 6 to 8 hours I still get top speeds, but it takes about 10 to 20 seconds to connect to anything. Half of the time I get an timeout and "page not found". Can't even update my linux boxes, as APT often can't find/resolve the configured repositories. Sometimes it takes just a lot longer to resolve the host so it works in the end.
Has anyone seen something like that?
I even tried SmartDNS with the same result. Works great after a reboot. All pages are loaded almost instantaneously in about 1 to 2 seconds. After about 6 to 8 hours everything needs about 20 seconds. Pages actually still do load fast when the host is resolved, unless they have more linked content witch needs several refreshes to actually see the page fully loaded.
Restarting dnsmasq doesn't change a thing. Only after a reboot everything is snappy again for some time.
Today, I executed a couple of traceroutes to random addresses I have never ever visited. I can see that name resolution in fact does work very fast. I get the IP back with the stroke of the enter key. It seems the router is somehow struggling with the traffic. After a reboot everything is fine again.
Today I did a reset and configured everything beside my vlans. Will see tomorrow if the connection gets degraded again. I also didn't change SFE to CTF, but have left it on default.
I'm also considering to switch to Fresh Tomato, since my router model is supported there too and check if the issue disappears.
So far I never had that of a big issue in years. Kinda interested in the cause of that behavior.
Joined: 18 Mar 2014 Posts: 13870 Location: Netherlands
Posted: Thu Aug 28, 2025 14:07 Post subject:
bermudahonk wrote:
Today, I executed a couple of traceroutes to random addresses I have never ever visited. I can see that name resolution in fact does work very fast. I get the IP back with the stroke of the enter key. It seems the router is somehow struggling with the traffic. After a reboot everything is fine again.
Today I did a reset and configured everything beside my vlans. Will see tomorrow if the connection gets degraded again. I also didn't change SFE to CTF, but have left it on default.
I'm also considering to switch to Fresh Tomato, since my router model is supported there too and check if the issue disappears.
So far I never had that of a big issue in years. Kinda interested in the cause of that behavior.
If you notice the slow down is there anything in the logs?
You might have exhausted your connections
Oh and I would not use SFE, either do not use any offloading or use CTF&FA (if your router supports it, the very early AC68 do not seem to support CTF&FA)
If you notice the slow down is there anything in the logs?
No, there was noting there besides the usual, mostly DHCP stuff.
Quote:
You might have exhausted your connections
Not a chance. I have 2 PCs I use separately. My desktop pc (LAN) and my notebook (WLAN). There ist about zero traffic and my connections are below 200. This setup has worked for years. I travel a lot so I access my network at home via VPN a lot.
Quote:
Oh and I would not use SFE, either do not use any offloading or use CTF&FA (if your router supports it, the very early AC68 do not seem to support CTF&FA)
My model is revision A2. It supports CTF & FA. But since I use QoS (mostly for bigger downloads/uploads), I have had CTF configured by default.
Thanks for your reply. I'm still doing some test setups. Just want to figure out what the culprit is. First I assumed I had trouble with my line. But everything was fine after I rebooted my router. But after several hours things slow down significantly again. By switching to FT I wanted to check if this is maybe somehow related to the router itself or my ISP.
my tests didn't really show anything substantial. Somehow my internet connection is simply randomly responding with lags after some time. Today, I still got some 404, but after hitting reload the page actually showed up. Interestingly using wget, I got "connection reset by peer" a couple of times too. This clearly indicates that my connection is dropped due to some errors on my side.
Anyway, I think this is somehow related to my WAN. Reboots seem to "refresh" the connection. So when things slowed down again, I simply did a DHCP renew on the WAN and I was able to connect to any random web-page within the usual 1 to 2 seconds. Some pages really appear fully loaded within a second, some load the structure and the content follows in about 2 to 3 seconds later.
I also tested my router tethered to my phone on 5G, and the result was, that there were no such long lags, but maybe the usual 3 to 4 seconds I would expect from mobile data as the providers don't do constant transmissions but bursts of data within small time frames.
Seeing this my issue points to my ISP.
Hello again,
after further investigation I was able to pinpoint the problem. It is now in fact the SFE option. By disabling it (I use mainly CTF) the issue did not appear at all anymore. Enabling it again caused the slowdown after several hours again. Restarting it via service call "healed" the connection instantly until it worsened again.
Can't really tell what is wrong with SFE in my Asus router. I have never had a case like that. I guess updating from a 2022 image to a 2025 image was a big leap forward and things are simply different. So far I have been running my current image without CTF with no other issues. I'm on 250MBit cable. As long as I can download at least with 10MB/s I simply don't really care about the speed. This is enough for all my purposes. I usually get around 15-20MB/s when doing updates anyway.
Joined: 16 Mar 2019 Posts: 413 Location: Szczecin, Poland EU
Posted: Thu Sep 25, 2025 11:41 Post subject:
It's simple question - Why you don't use build r62170 ? Update your device step by step. Currently you have an old build from May. It's look like some modules of your dd-wrt become overloaded. You can check the problem is on only DNS service or on whole router. What's going on when you manual restart Dnscrypt, dnsmasq services in command line. For example I'm use Smartdns module instead dnsmasq. I think Dnscrypt can be rather heavy for device (router).
Sorry, I can't spend endless time to switch images, since my config is very detailed. I had an issue and was mostly focused on pin-pointing the culprit, so I don't end up barking up the wrong tree. That is what I did. This issue is persistent BTW. All I do is update the topic.
So far things are running again. I will probably leave that and try a newer update next year. Others don't seem to have that issue. I can only confirm that disabling CTF has helped.
I also switched to SmartDNS and used my ISPs DNS with zero encryption, but that wasn't the problem in the end.
DNScrypt is not heavy at all. I use wireguard with several clients and even that does not make a real impact.
I never had any real performance issues with my router. It has 256MB RAM and a dual-core ARM @1.2GHz.
Nice to see someone from Szczecin in here. I was born there
