What's new

Release Asuswrt-Merlin 388.2 is now available for select models

  • SNBForums Code of Conduct

    SNBForums is a community for everyone, no matter what their level of experience.

    Please be tolerant and patient of others, especially newcomers. We are all here to share and learn!

    The rules are simple: Be patient, be nice, be helpful or be gone!

Status
Not open for further replies.
AX88U. Only thing not working for me (also in the beta's) is the internet-speedtest - it does test the speed ok, but cannot complete the whole test.
speed.png


Works perfectly here.

BTW uptime:
uptime.JPG

Without any trouble...

Thanks again @RMerlin
 
Not sure if it's related to the arp issue I've seen in other posts but after the upgrade my dual gt-ax6000s could not be on the same subnet without bringing everything down. If I pinged both at the same time while one responds the other times out and visa versa about every ten pings or so. I was hoping the update would fix the previous bug in 388.1 where I could not switch from router to AP mode and have it stay there - only a couple GUI changes and it would snap back to router mode. Upgraded router to Opnsense appliance thus the need for AP mode. Unfortunately I didn't have the luxury of testing that since these dead spiders needed to be brought back to life for some critical work asap. So unfortunately had to flash back to stock FW for the time being.
 
Asus AX11000 Pro, works fine for me.
Still the 5GHz LED's remain on even when the radio is disabled.
I have not updated to this Merlin release but I can say that the latest official and beta Asus firmware’s cause the same issue, 5ghz remains on when all radios disabled.
 
no issues to report
 
Sometime around midnight there was a bunch of topology changes captured on the scribe/ethernet log on the AX88U. From that moment going forward, the speedtest results went from (down/up) 930/930 to 400-650 / in one direction 500-750 in the other and a few packet drops. At first I though it was the scripts, then IPv6 (still can't get it to work like I had it in 386, anyone on AT&T Fiber shoot me a msg on how you have it working and how) that causes this. But the scripts, specifically scribe, we're reacting to the event, and IPv6 as I was going through trying to make it work likely also triggered it, whatever "it" is.

As a result of the topology changes, I've disabled Spanning Tree under LAN / Switch control. It's frustrating that by the time I noticed it, too much time had passed to capture it in the system log. I did manage to see the 30m window where it failed based on the speedtest results and correlate that to all the topology changes (ethernet.log / scribe). But I don't have what the trigger was that caused it. Turning off Spanning Tree obviously triggers a reboot, and as a result back to normal on the speedtest results.

Under 384/386 families never had anything like this. Something is off, but not enough to make it immediately noticible when it triggers. It's not bad enough where the wife is going to let me know that the Internet is out because she can't stream her telenovela, if she did at least that would warn me to get copies of the logs. 😣

Normally, just see a bunch of:
Router kernel: br0: received packet on eth7 with own address as source address

Even after it settles down those repeat but only a rebbot seems to put it back to my normal and the 930/930 Mbps results return
Scribe/Ethernet log attached...

:mad: Just going to have check the log(s) more often...
 

Attachments

  • scribe ethernet.txt
    7.1 KB · Views: 34
This speedtest shortens your life, @aex.perez. :eek:
Only when the packet losses that accompany it prevents wifey watching her telenovelas 🤣

The drop with the results and packet losses are a result of whatever is going on. Now all I need to do it capture it happening.

At this point considering turning one of the laptops into a syslog server to have plenty of capacity and detail for the router and nodes…
 
If you remove spdMerlin and stop hammering constantly your ISP connection for no reason your wife may be happier. You may even move out of the couch. The most Internet traffic generated by your household is perhaps to speedtest servers. Let your family members use some of this Internet connection.
 
Last edited:
I guess more details are in order, the bandwidth isn’t being hammered. Speedtest run as they always have, at 12 and 42 minutes of the hour. I got 1Gb fiber and when everything runs right the results are ~930Mbs, and always has been.

But when whatever is happening happens the first place that shows something is happening is Speedtest whether SPDMerlin on a schedule or not or using the the ASUS one.

There is no way I’m driving the router that hard. My CPU utilization across all CPU’s would be a constant 100% if I was consuming that much bandwidth. But once it starts, WiFi is next, followed by wired connections.

Not enough to outright fail but enough to see DCD tainted (not running anything TrendMicro) for dnsmasq and others and all sorts of weirdness that I’m not used to seeing in the logs, normally (my normal).

Thought the router was slowly dying but this is occurring only on 388, thought it was the carrier but the Speedtest on the modem are fine and no dropped packets.

Speedtests are not the problem. But it’s the best barometer I have in my tool chest at the moment to tell me that something broke and more or less when within a 30min window.

Next up, getting one of the PC’s running syslog and getting the router and nodes to send everything to that PC once I set the log levels (nvram) to capture as much detail as possible to be able to correlate it all and hopefully see what’s broken/breaking.

Once it triggers the only way to fix it is a reboot and like a light switch the results go back to normal and no lost packets.

For now Speedtests is my canary in the coal mine so to speak. Let’s me know there’s carbon dioxide in the mine that can kill me, but not enough to tell me from exactly where or when…

Any tools/ideas that can help me to capture what happens when it happens, suggestions where to look on the router/nodes, I’m listening.

But let me clear it up, not complaining about Speedtest results
 
Any tools/ideas that can help me to capture what happens when it happens

Remove all the scripts and test the firmware first. Add only what's needed. You are stats, graphs and speedtests obsessed and they do affect your entire family's user experience. Twice per hour your ISP line is saturated and everything else slows down so you can have a graph. spdMerlin and connmon run there for your satisfaction. scMerlin is a lazy boy tool, basically. What you run ntpMerlin for? YazDHCP - easy replacement, just copy/paste 3 files with SCP and done. Skynet and scribe/uiscribe just in case or actually doing something? Pushing for 160MHz and DFS plus SmartConnect? Why?

I have my test AX86U set with clean 388.2 install, no TrendMicro, Cake QoS to stress the CPU a bit, AdGuard Home with 10 blocklists, 4 allowlists, 2x DoH upstream servers, Entware bacause it's needed, 2GB swap and added Skynet because many are using it with or without reason. The router runs fine with no errors in logs and about 400MB free RAM. Speedtest constant at what the CPU allows with Cake and A on popular (and buggy) bufferbloat sites. I tested with all TrendMicro enabled as well and still good. What else is needed? Unbound - I've tried that too, works fine, no errors in logs.

Sorry @aex.perez, but I start to believe you create the issue first and then fight with it.
 
Done the dirty from Beta 2 about an hour ago. Have done a full system restart and will let it settle.

Initial upgrade was fine. No issues observed so far.
- DDNS good
- DNS Director good
- VPN Director good
- IPTV good
- Nest gear and HD cams good
- IOT/TP-Link Devices good
- Internet speed as expected

View attachment 49272

Let's GO!

Cheers again @RMerlin and team!
I concur

1681710243358.png
 
Remove all the scripts and test the firmware first. Add only what's needed. You are stats, graphs and speedtests obsessed and they do affect your entire family's user experience. Twice per hour your ISP line is saturated and everything else slows down so you can have a graph. spdMerlin and connmon run there for your satisfaction. scMerlin is a lazy boy tool, basically. What you run ntpMerlin for? YazDHCP - easy replacement, just copy/paste 3 files with SCP and done. Skynet and scribe/uiscribe just in case or actually doing something? Pushing for 160MHz and DFS plus SmartConnect? Why?

I have my test AX86U set with clean 388.2 install, no TrendMicro, Cake QoS to stress the CPU a bit, AdGuard Home with 10 blocklists, 4 allowlists, 2x DoH upstream servers, Entware bacause it's needed, 2GB swap and added Skynet because many are using it with or without reason. The router runs fine with no errors in logs and about 400MB free RAM. Speedtest constant at what the CPU allows with Cake and A on popular (and buggy) bufferbloat sites. I tested with all TrendMicro enabled as well and still good. What else is needed? Unbound - I've tried that too, works fine, no errors in logs.

Sorry @aex.perez, but I start to believe you create the issue first and then fight with it.
Already been down that road. Did all that and still nothing that I can point a finger to as a root cause. Let me be clear, this has been 100% with 386. And as long as whatever happens doesn't happen in 388, it's 100% as well don't assume that I'm blasting the router with speedtests. But without knowing what the cause is and what cases it, the only way to see that something happened is from the speedtest results. With respect to the speedtests the results correlate 100% to the start and stop (post reboot). So take a seat in the back of the room unless you got something more constructive to that'll help pinpoint the issue. Or reread the post to better understand what's been done to find the issue. The last change is Spanning tree off, might has to reverse that. I know with IPv6, under certain conditions I can cause it, but with the logs being flodded and jumping between screens its too hard to correlate.

Let me get the router and nodes going to a common syslog (with lots of capacity) for correlation and we'll see what we see.
 
Already been down that road. Did all that and still nothing that I can point a finger to as a root cause. Let me be clear, this has been 100% with 386. And as long as whatever happens doesn't happen in 388, it's 100% as well don't assume that I'm blasting the router with speedtests. But without knowing what the cause is and what cases it, the only way to see that something happened is from the speedtest results. With respect to the speedtests the results correlate 100% to the start and stop (post reboot). So take a seat in the back of the room unless you got something more constructive to that'll help pinpoint the issue. Or reread the post to better understand what's been done to find the issue. The last change is Spanning tree off, might has to reverse that. I know with IPv6, under certain conditions I can cause it, but with the logs being flodded and jumping between screens its too hard to correlate.

Let me get the router and nodes going to a common syslog (with lots of capacity) for correlation and we'll see what we see.
Turn any "extra" features off and you will be going faster than ever!
 
From my own experience, I do tend to agree with @Tech9 in principle - basically a 'less is more' approach with the router. Why? It's because I want maximum reliability and consistency (keep wifey and kids happy). Simple as that.

So I started from scratch with my RT-AX86S and do almost no changes from default merlin. No swap, no usb - nothing apart from cake, copy the static dhcp files and make a few tweaks to wifi settings. Add my Aimesh nodes. Done.

I do have adguard and unbound on a separate device (RPI3/4) and this is great too - one less thing to setup and maintain on the router. It is just there doing its thing and clients use it through their dhcp settings.

In terms of testing network stuff. I do that as needed on client devices by running iperf on a server wired to the router. This helps isolate LAN/wifi issues on the network - which is usually the route issue TBH. From these same client devices I also test internet speed and buffer bloat occasionally.
 
Turn any "extra" features off and you will be going faster than ever!
Geez, how many times 🤦🏻‍♂️

It’s not the SpeedTest results I’m worried about. It’s what’s causing them and all the reboots and instability with packet drops, dnsmasq, and so on.

Tell you what, once I got centralized syslog going, then I can do a HW reset again and start from scratch and it’ll still happen. But even if it doesn’t as often or at all, the simple fact that this same setup has been flawless during all the 386 releases, should be enough to merit a deeper dive. Unfortunately today my only way to know is when Speedtest results go bonkers. I’ve got no way to identify any issues between scrips or with Entware that could be contributing to this. Nor do I have a testing/debugging setup to deep dive into this or I would have found it by now and mitigated it.

Tomorrow that deeper dive begins once I get all the logs in the same place. If you have any recommendations for tools or where to look for anomalies please share.
 
Geez, how many times 🤦🏻‍♂️

It’s not the SpeedTest results I’m worried about. It’s what’s causing them and all the reboots and instability with packet drops, dnsmasq, and so on.

Tell you what, once I got centralized syslog going, then I can do a HW reset again and start from scratch and it’ll still happen. But even if it doesn’t as often or at all, the simple fact that this same setup has been flawless during all the 386 releases, should be enough to merit a deeper dive. Unfortunately today my only way to know is when Speedtest results go bonkers. I’ve got no way to identify any issues between scrips or with Entware that could be contributing to this. Nor do I have a testing/debugging setup to deep dive into this or I would have found it by now and mitigated it.

Tomorrow that deeper dive begins once I get all the logs in the same place. If you have any recommendations for tools or where to look for anomalies please share.
A quick look at the ole' google came up with this

 
the simple fact that this same setup has been flawless during all the 386 releases

As far as I remember the author of all "j" scripts moved to pfSense appliance after his Asus router died. When the scripts you use were last updated? They don't even know 388 firmware exists. Perhaps this is your difference between 386 and 388 installation. Five of your scripts are monitoring something or sorting log data. How many times you login to the router's GUI daily and why the need to monitor so much a Fiber connection? If you run spdMerlin twice per hour you generate >14GB traffic only in speedtests daily. Also I've noticed in my tests the sequence you install those scripts may alter the outcome. One of them wiped the custom configuration I had with no questions asked. Took me some time to figure out what's going on.

My CPU utilization across all CPU’s would be a constant 100% if I was consuming that much bandwidth.

This is not how it works. All routing is done on Core 1. Even if the core is not loaded 100% your ISP line is. Close to 100% of available bandwidth is taken by the speedtest script. It loads the CPU and needs RAM. Your router may start swapping on the dead slow USB swap file. During that time you may observe significantly increased latency and dropped packets. Your wife may see lags and freezes with streaming services. Some services may crash with OOM condition. If you have USB attached storage - it makes it worse. This is a home router with limited resources. What you report in 388 firmware test/release threads may not be related to the firmware at all. If no one else with AX88U is having similar issue - it's somewhere around your configuration.
 
Hello guys,
In my case, I am still on the 386.8 firmware. I saw lately the possibility of ARP errors or other problems since the new GPL of 388.X versions...
So I'll wait a bit more to update my router.
With the 386.8, no problem since almost 250 days, that is 8 months. LAN/WAN/WIFI solid as a rock of Jupiter 🚀
Thanks again to Merlin and the Developers ;)
1681732727844.png

1681732736645.png
 
Last edited:
Status
Not open for further replies.

Similar threads

Support SNBForums w/ Amazon

If you'd like to support SNBForums, just use this link and buy anything on Amazon. Thanks!

Sign Up For SNBForums Daily Digest

Get an update of what's new every day delivered to your mailbox. Sign up here!
Top