FlexQoS FlexQoS issues with 388.4 HND5.04 models

[dave14305]

Thank you for your commitment and I look forward to hearing from you. Kind regards.

When it’s a convenient time to test again and risk a reboot, please add back the WiFi Calling rule, but amend it to add the Mark 0b0042. Then make a call. The theory is that with a mark filter on the rule, it will not act on the packets once their mark changes to that 0x1… value. And hopefully not reboot.

 

[dexu]

When it’s a convenient time to test again and risk a reboot, please add back the WiFi Calling rule, but amend it to add the Mark 0b0042. Then make a call. The theory is that with a mark filter on the rule, it will not act on the packets once their mark changes to that 0x1… value. And hopefully not reboot.

We have a genius on this forum I set it to 0b0042 and the router did not reset! I leave it, it's worth adding it as default

 

[dave14305]

We have a genius on this forum I set it to 0b0042 and the router did not reset! I leave it, it's worth adding it as default

If you add the logging rule again and stay on the call long enough, does the mark in the log ever change back to the original value?

I appreciate your cooperation. I will publish a workaround to filter on only “valid” marks for now, until we understand the problem better.

 

[dave14305]

I’ve pushed v1.3.4 to the develop branch just to make sure it works for others before pushing to stable, since I don’t have an Asus router to test myself. If you are brave, please switch by running flexqos develop.

This is just a workaround for now. Thanks to everyone who was able to provide crashdumps and logs.

 

[TurnstileBlues]

I’ve pushed v1.3.4 to the develop branch just to make sure it works for others before pushing to stable, since I don’t have an Asus router to test myself. If you are brave, please switch by running flexqos develop.

This is just a workaround for now. Thanks to everyone who was able to provide crashdumps and logs.

I'm feeling brave

I'll try on my 86U-pro

 

[Yiannis]

alright , just installed and switched to develop branch.

 

[TurnstileBlues]

I’ve pushed v1.3.4 to the develop branch just to make sure it works for others before pushing to stable, since I don’t have an Asus router to test myself. If you are brave, please switch by running flexqos develop.

This is just a workaround for now. Thanks to everyone who was able to provide crashdumps and logs.

This is working just fine for me now!

I'll put the router under a bit of pressure to see what happens. But everything appears to be good at the moment

 

[dexu]

If you add the logging rule again and stay on the call long enough, does the mark in the log ever change back to the original value?

I appreciate your cooperation. I will publish a workaround to filter on only “valid” marks for now, until we understand the problem better.

There is only one entry, strangely enough, when I dial another number again, the next one does not appear.

Sep 21 22:01:48 kernel: [FlexQoS Debug] IN= OUT=eth0 SRC=192.168.2.253 DST=89.108.200.112 LEN=180 TOS=0x18 PREC=0xA0 TTL=63 ID=24952 PROTO=UDP SPT=42412 DPT=4500 LEN=160 MARK=0xf578180

Mark 0b0042 is set... and he generated a different one.

 

[dave14305]

There is only one entry, strangely enough, when I dial another number again, the next one does not appear.

Sep 21 22:01:48 kernel: [FlexQoS Debug] IN= OUT=eth0 SRC=192.168.2.253 DST=89.108.200.112 LEN=180 TOS=0x18 PREC=0xA0 TTL=63 ID=24952 PROTO=UDP SPT=42412 DPT=4500 LEN=160 MARK=0xf578180

Mark 0b0042 is set... and he generated a different one.

I will pretend I never asked, since this new data is more confusing. Instead, I will enjoy the small victory for today!

I went ahead and pushed the 1.3.4 release to stable branch. It’s fairly simple so I think the risks are low.

 

[Yiannis]

Hi @dave14305 it looks the latest version is stable! Many thanks! Just out of curiosity, you mentioned you don't own an ASUS router?
So, you are not using FLexQoS or does it run to other platforms too?

 

[dave14305]

Hi @dave14305 it looks the latest version is stable! Many thanks! Just out of curiosity, you mentioned you don't own an ASUS router?
So, you are not using FLexQoS or does it run to other platforms too?

My home network is now based on OpenWrt. I have a nice CAKE setup with some minimal diffserv3 marking using nftables, dnsmasq nftset integration and tc-ctinfo.

Technically I still own an Asus router, but it’s in storage.

 

[Kingp1n]

My home network is now based on OpenWrt. I have a nice CAKE setup with some minimal diffserv3 marking using nftables, dnsmasq nftset integration and tc-ctinfo.

Technically I still own an Asus router, but it’s in storage.

@dave14305 Is the openwrt cake similar to the built-in cake in rmerlin firmware?

 

[dave14305]

@dave14305 Is the openwrt cake similar to the built-in cake in rmerlin firmware?

CAKE is the same, more or less, but the benefits come from the rest of the more modern software stack. For example, I have CAKE working pretty well with nftables software flow offloading (what we hear would call NAT acceleration). And I only have to worry about assigning DSCP marks once on egress, knowing that the tc ctinfo module will restore the same DSCP on the return ingress traffic.

The grass isn’t always greener, but it’s a nice alternative.

 

[dexu]

I will pretend I never asked, since this new data is more confusing. Instead, I will enjoy the small victory for today!

I went ahead and pushed the 1.3.4 release to stable branch. It’s fairly simple so I think the risks are low.

Updated and working! Best Regards Dave.

 

[dave14305]

Updated and working! Best Regards Dave.

When you have a moment, this week ahead, please open an ssh session on the router and run tc -ts monitor. Then make a call and see if any messages get logged in the ssh window for a minute. If not, ctrl-c out of the monitor command and carry on.

I’m wondering if any filters get created to address these odd marks.

It would also help to share the output of tc filter show dev br0 to see if anything new is in 388.4.

Thanks!

 

[dexu]

When you have a moment, this week ahead, please open an ssh session on the router and run tc -ts monitor. Then make a call and see if any messages get logged in the ssh window for a minute. If not, ctrl-c out of the monitor command and carry on.

I’m wondering if any filters get created to address these odd marks.

It would also help to share the output of tc filter show dev br0 to see if anything new is in 388.4.

Thanks!

Hi. tc -ts monitor shows nothing during conversation.. peace and quiet.
tc filter show dev br0 -> https://pastebin.com/NpX9XAse

Best Regards, Marcin.

 

[dave14305]

Hi. tc -ts monitor shows nothing during conversation.. peace and quiet.
tc filter show dev br0 -> https://pastebin.com/NpX9XAse

Best Regards, Marcin.

Thanks, unfortunately we don’t learn much then. Is this with fq_codel mode in FlexQoS? How about trying again in “ASUS“ mode?

 

[dexu]

Thanks, unfortunately we don’t learn much then. Is this with fq_codel mode in FlexQoS? How about trying again in “ASUS“ mode?

Same here ..
Catch https://pastebin.com/9hyehT4n

M.

 

[dave14305]

Same here ..
Catch https://pastebin.com/9hyehT4n

M.

Thanks. I was able to reproduce the original problem on an RT-AX86U PRO, running stock firmware for now. I’ll see if I can dig deeper.

 

[Kingp1n]

Thanks. I was able to reproduce the original problem on an RT-AX86U PRO, running stock firmware for now. I’ll see if I can dig deeper.

I see you got your router out from storage? Hope you can get more info on this!!!