[ 386.4 alpha Build(s) ] Testing available build(s)

[SomeWhereOverTheRainBow]

There are some odd regressions being discussed on the dnsmasq-discuss list from the PiHole guy (Dominik).

[Dnsmasq-discuss] [BUG] [PATCH] Segmentation fault in src/forward.c

[Dnsmasq-discuss] change in behavior where v4 address exists but not v6 in 2.86

I see Merlin is testing 2.86 out with a reversion to nettle instead of openssl since theMiron isn’t around as much anymore to maintain his adaptation.

It looks like themiron changelog already includes the changes. Would Rmerlin have to just port them over? Or is theMiron missing parts of the version changes?

Merge remote-tracking branch 'upstream/master' into openssl · themiron/dnsmasq@989ee98

Dnsmasq DNS proxy and DHCP server with OpenSSL backend - Merge remote-tracking branch 'upstream/master' into openssl · themiron/dnsmasq@989ee98

github.com

The pihole regressions issues are related to their own customized dnsmasq and related configurations.

 

[RMerlin]

Themiron's repo hasn't been updated since early April. It's missing the vast majority of 2.86 commits, and these conflict with the OpenSSL patches, hence I'm looking into switching back to vanilla + nettle for the future, so I can more easily update whenever dnsmasq issues critical security updates, for example.

 

[SomeWhereOverTheRainBow]

Themiron's repo hasn't been updated since early April. It's missing the vast majority of 2.86 commits, and these conflict with the OpenSSL patches, hence I'm looking into switching back to vanilla + nettle for the future, so I can more easily update whenever dnsmasq issues critical security updates, for example.

why did you ever move from nettle to openssl any ways and hopefully it wont break anything going back.

 

[bradbort]

I did a dirty upgrade from Merlin's latest to alpha on an AX88U. Seemed to run fine, but the list of wifi clients was a bit fargled up. It did not have them correctly listed by 2.4 vs 5 vs direct, and speeds were absent.

 

[RMerlin]

why did you ever move from nettle to openssl any ways and hopefully it wont break anything going back.

1) One less security-related library to maintain and keep up-to-date (since OpenSSL was already there)
2) OpenSSL supported a few ciphers that nettle didn't
3) OpenSSL is more tightly optimized on ARM
4) The plan was to have OpenSSL support merged upstream by the dnsmasq author, but ultimately he showed little interest in doing so

 

[SomeWhereOverTheRainBow]

1) One less security-related library to maintain and keep up-to-date (since OpenSSL was already there)
2) OpenSSL supported a few ciphers that nettle didn't
3) OpenSSL is more tightly optimized on ARM
4) The plan was to have OpenSSL support merged upstream by the dnsmasq author, but ultimately he showed little interest in doing so

Sad it never got any further, but you gave it a good go.

 

[capncybo]

I have a brand new 4 day old RT-AX56U and it crashed while running Asus Merlin 386.4 alpha1. I am running no scripts and have a pretty simple configuration as in my signature (2 guest networks using Guest Network 1 on each band, 2 main networks, DNS over TLS, IPv6).

Prior to the crash, I was streaming a movie on the 5Ghz guest network. CPU load is very low always on this router and CPU temp runs 50 - 60 degrees C. After the crash, 2 Wyze cameras that I have could no longer connect to the 5Ghz guest network despite power cycling the cameras and other troubleshooting. I rolled back to 386.3.2 to be able to connect those devices again.

I would appreciate someone with more knowledge than I have to dissect the crash log as to what may have happened.

Ummm I don't know if I have more knowledge than you but...
As a fairly new member of the WYZE user club:
The WYZE cams do NOT support 5GHz
As they only ever communicate via 2.4GHz.
However, you can use 5GHz on any client to access the stream(s) through your router.

EDIT: doesn't explain your router crash, but it does explain their so called failure to connect @5G WiFi.

 

[JWoo]

I just misspoke. They are on 2.4 you are correct.

 

[ItsMeDude]

I did a dirty update from 386.3_2 on an AX88U with an AX68U node, and they've been solid for about 10 days. No issues at all.

 

[Kingp1n]

Anyone here upgraded to the latest alpha can confirm if this alpha version include the fixes from the recent official firmware (3.0.0.4.386.45375/ released 1 Sep 2021) for the GT-AX11000? I appreciate the info.

 

[octopus]

Anyone here upgraded to the latest alpha can confirm if this alpha version include the fixes from the recent official firmware (3.0.0.4.386.45375/ released 1 Sep 2021) for the GT-AX11000? I appreciate the info.

386.2 (2-Apr-2021)
- UPDATED: Merged GPL 386_42095.

 

[L&LD]

Report: 2x RT-AX68U in AiMesh (wireless backhaul) mode with v386.4 Alpha 1 RMerlin firmware.

Background: Customer has upgraded to nominal 600Mbps down, 100Mbps up, speeds (from 300/50Mbps) ISP is capable of sustained 650d/110u performance. Upgraded from 2x RT-AC86U in AiMesh wireless backhaul mode with identical firmware. Home on four levels and approximately 3,900 SqFt of usable area.

Executive summary: As expected, the newer RT-AX68U's excelled at throughput vs. the older design RT-AC86U's. Actually, surprisingly so. While the RT-AC86U could hit 150-210 Mbps download to the lowest level from the AiMesh node a floor and many walls above. The RT-AX86U could hit 350-430Mbps with a single router (before the AiMesh node was added) to this same device (current laptop, plugged into AC power) from two floors above and a few extra walls in between too.

Before installing at the customer's home, I tried to get the RT-AX68U to be added as a node to my (two) RT-AX86U's. It was running a very old 386.xx stock firmware (386.4069x, from memory). It could connect but would continually reset itself and effectively make the network unusable. Flashing to RMerlin 386.4 Alpha 1 made it correctly work with my RT-AX86U's without further issues on the wireless front.

On an interesting note, after I saw the performance was above my expectations on my network, I tried using the WPS Button reset method for the RT-AX68U and could not successfully reset it. Using the GUI wasn't an issue (or the Reset Button itself). Maybe @RMerlin needs to investigate this? And of course, it may have been already fixed in the latest GPL releases for this model.

While a single RT-AX68U proved very capable of replacing a 2x RT-AC86U setup, the customer insisted on having a second RT-AX68U installed as an AiMesh node. Again, I tested with the shipping firmware noted above, and the WPS Button reset worked every time. The node also connected immediately to the main router too with the stock firmware. After upgrading to the 386.4 Alpha 1 firmware, the WPS Button reset would not work on the node either.

With a beefy gaming desktop and an Intel AX200 Wi-Fi adaptor onboard through a couple of walls on the same floor as the main router, the speeds could easily hit the ISP's maximums (650/110). With random peaks (thanks to cable modems!) to 800Mbps or even higher).

For (nominally) $60 cheaper than an RT-AX86U, the RT-AX68U is a good match for less than 1Gbps service. I will be keeping a close eye on this installation to see if the performance and stability continue.

Compared to the slightly less expensive RT-AX58U (around $10 less here), the RT-AX68U is the better router by far.

Except for the WPS Button reset not working with this Alpha 1 build, the router has shown to be a very worthwhile upgrade over the previously recommended class leading AC-class RT-AC86U.

I don't recall reading about any WPS Button issues with this model and this Alpha 1 firmware? Should I try yet another example to test for hardware issues? (The performance and stability of the setup in the last week do not point to any such conclusions so far).

 

[JWoo]

Report: 2x RT-AX68U in AiMesh (wireless backhaul) mode with v386.4 Alpha 1 RMerlin firmware.

Background: Customer has upgraded to nominal 600Mbps down, 100Mbps up, speeds (from 300/50Mbps) ISP is capable of sustained 650d/110u performance. Upgraded from 2x RT-AC86U in AiMesh wireless backhaul mode with identical firmware. Home on four levels and approximately 3,900 SqFt of usable area.

Executive summary: As expected, the newer RT-AX68U's excelled at throughput vs. the older design RT-AC86U's. Actually, surprisingly so. While the RT-AC86U could hit 150-210 Mbps download to the lowest level from the AiMesh node a floor and many walls above. The RT-AX86U could hit 350-430Mbps with a single router (before the AiMesh node was added) to this same device (current laptop, plugged into AC power) from two floors above and a few extra walls in between too.

Before installing at the customer's home, I tried to get the RT-AX68U to be added as a node to my (two) RT-AX86U's. It was running a very old 386.xx stock firmware (386.4069x, from memory). It could connect but would continually reset itself and effectively make the network unusable. Flashing to RMerlin 386.4 Alpha 1 made it correctly work with my RT-AX86U's without further issues on the wireless front.

On an interesting note, after I saw the performance was above my expectations on my network, I tried using the WPS Button reset method for the RT-AX68U and could not successfully reset it. Using the GUI wasn't an issue (or the Reset Button itself). Maybe @RMerlin needs to investigate this? And of course, it may have been already fixed in the latest GPL releases for this model.

While a single RT-AX68U proved very capable of replacing a 2x RT-AC86U setup, the customer insisted on having a second RT-AX68U installed as an AiMesh node. Again, I tested with the shipping firmware noted above, and the WPS Button reset worked every time. The node also connected immediately to the main router too with the stock firmware. After upgrading to the 386.4 Alpha 1 firmware, the WPS Button reset would not work on the node either.

With a beefy gaming desktop and an Intel AX200 Wi-Fi adaptor onboard through a couple of walls on the same floor as the main router, the speeds could easily hit the ISP's maximums (650/110). With random peaks (thanks to cable modems!) to 800Mbps or even higher).

For (nominally) $60 cheaper than an RT-AX86U, the RT-AX68U is a good match for less than 1Gbps service. I will be keeping a close eye on this installation to see if the performance and stability continue.

Compared to the slightly less expensive RT-AX58U (around $10 less here), the RT-AX68U is the better router by far.

Except for the WPS Button reset not working with this Alpha 1 build, the router has shown to be a very worthwhile upgrade over the previously recommended class leading AC-class RT-AC86U.

I don't recall reading about any WPS Button issues with this model and this Alpha 1 firmware? Should I try yet another example to test for hardware issues? (The performance and stability of the setup in the last week do not point to any such conclusions so far).

What kind of CPU and wireless temperatures are you seeing a) during low load and b) during high load? What kind of CPU loads do you see?

 

[maxbraketorque]

Report: 2x RT-AX68U in AiMesh (wireless backhaul) mode with v386.4 Alpha 1 RMerlin firmware.

....

Executive summary: As expected, the newer RT-AX68U's excelled at throughput vs. the older design RT-AC86U's. Actually, surprisingly so. While the RT-AC86U could hit 150-210 Mbps download to the lowest level from the AiMesh node a floor and many walls above. The RT-AX86U could hit 350-430Mbps with a single router (before the AiMesh node was added) to this same device (current laptop, plugged into AC power) from two floors above and a few extra walls in between too.

...

In a sense, this may not be a surprising result because the AC86U is only a dual band router, so in wireless backhaul, the 2.4 GHz band has to be utilized either for the backhaul or the connection to the customer device. Having said that, was customer device wifi communication protocol the same for these comparisons? I could see the single AX68U setup further outperforming if it was using AX to talk to the customer device.

Did you repeat the test with dual AX68U's in AiMesh mode?

 

[miniterror]

Report: 2x RT-AX68U in AiMesh (wireless backhaul) mode with v386.4 Alpha 1 RMerlin firmware.

I cant remember wich firmware i had installed but had the same problem on a AX68U, i could not get WPS reset working.
Also had some problems with it being a mesh node connected with wire to a AX86U, eventually returned it and went for a second AX86, since no more problems for me.
So my personal experience wasnt too good with this model, be sure to keep a eye out as you state.

Edit: just read back and i had problems that the next morning none of my clients would connect to the node.
I had a daily reboot schedule active so maybe it was related to the daily reboots but with the AX86 i dont have those problems..
It was running latest stable Merlin firmware and on that i definetly had the same none working WPS reset procedure.

 

[L&LD]

@JWoo, 62C at zero/low load (before I shared the new SSID passwords with the household). Before I left, it was still less than 74C at 42% CPU load with 18 active devices (mostly streaming, speed tests, and gaming) and the units 'on' for over 60 minutes. The RT-AC86U was at 82C when I arrived at the home (don't remember the CPU load).

@maxbraketorque, the RT-AX68U is also a dual-band router (as are the RT-AX88U and the RT-AX86U too). I was expecting higher throughput because of my experience with the RT-AX88U and the RT-AX86U, but having such an increase in throughput, through such obstacles and, at a distance is always refreshingly surprising. To reach the speeds I was seeing, the 5GHz band was being used as both backhaul and for client devices. Very impressive speed boosts (for the distances/obstacles within that environment).

Yes, the 2x RT-AX68U's were a direct replacement for the previous RT-AC86U's, set up in wireless backhaul AiMesh mode. And, in AiMesh mode, the performance was (on average) the same for the farthest computer in the lowest level of the home (connected to the node) as when the node was powered down and the connection was to the main router.

The 'beefy gaming desktop' was the only device in the home with an Intel AX200 adaptor card (besides my work computer with an Intel AX210 adaptor). This computer could max out the ISP speeds at will (as could my work computer too, of course).

The main improvements that the customer received with this router upgrade:

• Faster connection speeds to their networks for all devices. This was evidenced by the fact that when I was rebooting the routers, the users were sometimes not even aware that I had rebooted the network.
• Lower latency for all clients. Younger (the 'gamer' in the household), older, and non-techy users all commented on how much faster their devices felt on the 'net. Yes, it was that noticeable.
• Higher throughput, throughout the home, regardless of the device used. The AiMesh node was 'required', not for enhancing the internal Wi-Fi coverage, but for extending the Wi-Fi to the end of the lawn/garden too (about another 70' from the house, and around 110' from the main router).

I followed up again today with the customer, and the status of the network was still 'great'.

Also, I want to note that none of the defaults (after flashing RMerlin firmware and doing a full reset on both routers) were changed with regards to the AX capabilities. All of their clients connected instantly to the new SSIDs and all have stayed connected. A mix of iPhones, Androids, Surface, laptops, desktops, TVs, etc.

While the performance, latency, and throughput improvements are unquestionable from the old RT-AC86U, the RT-AX68U looks and feels cheaper than the older model. They both use the same power adaptor (and identical 19V, 1.75A, and 33W specs), and power plugs. (Yes, I tested the new power plug with the old router and the old power plug with the new router too).

If the WPS Button reset method gets fixed with RMerlin firmware (anyone with updated Asus stock firmware can confirm that it works correctly for them?), then the recommended routers today would be:

RT-AX68U (entry-level). RT-AX88U (AX class and greatest 2.5GHz range). RT-AX86U (best overall router).

The RT-AC86U was a good router for many years. But, it has been surpassed (particularly when it is only $20 less at regular prices, (and $50 more expensive when the RT-AX68U is on sale).

And for the record, I was wrong about the RT-AX68U. So far, it has exceeded my expectations (contrary to what I expected after I had seen the specifications page on it, not so long ago).

I hope I didn't miss any questions?

 

[jsbeddow]

@L&LD , just curious as to why you consistently refer to what is customarily called the 2.4 Ghz wifi band as the 2.5Ghz band? Yes, the channels do run between the two, but the standard/traditional nomenclature has always been to call it the 2.4Ghz band.

Regardless, thanks for the updates on the results of this upgrade.

 

[L&LD]

@jsbeddow you're welcome and thank you for the question (and the chance to explain myself). To be sure, there is no 2.5GHz band (the highest channel (14) is still centered at 2.477GHz (see the last link below).

Caution: my own 'definitions' follow, as related to my customers, in semi-layman's terms.

The 2.4GHz band moniker is for all AC-class and lower Wi-Fi devices. The pinnacle to me being the RT-AC3100 and previously, the RT-N66U, and originally, the WRT54G. The biggest indicator/benefit of those 2.4GHz AC-class wireless routers was the excellent range and usable connections at the fringe they could achieve.

The 2.5GHz band moniker that I use today is usually for the better, AX-class routers I have used. While the absolute range may not be the same as the routers indicated above (in the strict 2.4GHz band), they do offer higher throughput at more reasonable distances for almost all client devices, while also being more resistant to interference too (that is also part of the router design too, of course. For example, shielded USB 3.0 ports, etc.).

While they are technically the same band, I use 2.5GHz to indicate the newer AX-class routers. On these forums, I may have included the RT-AC86U in this class of devices too for the 2.4GHz band (but on these forums, unlike when I'm talking to a customer, I'm speaking about all routers I've used).

The following are interesting links and may indicate how non-Wi-Fi interference can suddenly appear, even for users in the countryside, in sparsely populated neighborhoods, or other seemingly remote locations with no 'logical' change in their Wi-Fi environments.

The History of WiFi: 1971 to Today - CableFree

Using Sure Cross Wireless Products in a 802.11 (Wi-Fi) Environment (bannerengineering.com)

How to Choose the Best Wi-Fi Channel for Your Network (lifewire.com)

 

[shabbs]

I call it the 245Ghz band.

 

[Amiga]

@L&LD The RT-AX68U is on sale for $160 USD from Amazon and Office Depot / OfficeMax. I've been debating about whether to swap one for a RT-AC86U and you've solidified my decision since I've been dealing with flaky 2.4GHz.

Curious if you're ever nervous about putting Merlin and a beta on a customer's device? I always try to flash other people's Asus routers with the latest official firmware so I don't get any grief down the road.