Beta Asuswrt-Merlin 386.1 Beta (stage 2) is now available

[sentinelvdx]

Run "top" to determine what is using the CPU.

Sadly I had to reboot the router, but happened a couple of times before. I'll pay attention to next time it happens to run the cmd and post it here. thanks!

 

[JohnSmith]

Upgrade RT-AX88U & RT-AC3100 from V384.19 Final to V386.1 Beta5 via dirty firmware upgrade. 30+ devices (includes gaming, streaming, downloads, etc.) and all appears to be working for main AX88U & backup AC3100 routers. Next steps will be to add RT-AX58U with V386.1 Beta5 into AiMesh 2.0, and see if guest networks and IOT network can be seperate but run across all of these AP's and routers seemlessly (fingers crossed).

 

[aex.perez]

AX88U HW 1.1 B4b to B5 via GUI / 2 AC5300 AImesh B4 to B5 via GUI.

3 "RT-AX88U kernel: DUMP CONSOLE:" events and 2 reboots (1 dump occurred during the soft/via GUI reboot)

The router just went nuts and started dropping client and services. First networks WiFi, both bands, LAN, WAN, GUI but not all clients, not all at once. Then other services (dnsmasq, GUI) and other scripts (scMerlin, spdMerlin, VPNmanagr, Skynet, ntpMerlin) dropped, again not all at once. CPUs got busy, as a result what clients were left on the network crawled to a stop with no internet access as WAN and dnsmasq dropped. Happened once while I was asleep and then at 19:30 all 1/28. I was able to reboot via the GUI as it came back briefly as a scMerlin restart. The router rebooted, hung and then Dump and rebooted again.

Here is the link to the log, starts on 1/25 to gather some history and then all the way to the recovery, after third dump/second reboot on 1/28 and a little further to 19:51 - Dropbox - AC88U dump/reboot log
Log was too long, even just one of the dump sections, couldn't post it any other way

I know what I'm doing this weekend, starting from a clean slate on all 3 on B5

 

[davidangel64]

@RMerlin
Everything is working fine so far on my RT-AX86U on Beta 5 except for the "Only Specified MACs" feature under TOR as it currently routes MAC addresses NOT specified to go through TOR.

 

[thecheapseats]

Sadly no. No power light no rescue mode.

hmmm... odd that the ax86 didn't do a sanity check on the image and stop the upload... but there has to be a way to 'fool' the router into taking a firmware upload...

if not, (and you have the skills) you can always open the router (void your warranty) and hook-into the serial TTL lines (tx, rx, gnd) on the inside of the unit and initiate a tftp session...

keep in mind, you MUST use a usb to TTL level serial cable - NOT a common usb to serial cable, which will trash the onboard ttl serial device and probably other stuff too... also do NOT connect the 4th pin (vcc) - just tx, rx and gnd... ftdi sells proto cables direct... ftdi p/n TTL-232-RPi or p/n TTL-232R-3V3 will work - the former being the easiest...

first thing I do to every new router I own is hook into the those serial lines and attach a b/t wireless (and wired) serial comm for out-of-band cli - I don't care about warranties...

 

[RMerlin]

hmmm... odd that the ax86 didn't do a sanity check on the image and stop the upload... but there has to be a way to 'fool' the router into taking a firmware upload...

Firmware images do get validated when being flashed. Both the actual model as well as a CRC32 integrity check to ensure the image isn't corrupted. Anyone who managed to flash an AX86U firmware on an AC86U must have done something really unusual to flash it, like using TFTP...

 

[joegreat]

Just to double-check, if I'm coming from 384_19_0 on my RT-AX88U, I flash 386.1 Beta 5 ... then do a Factory Reset?

If you plan to drive with your car on some muddy road (firmware upgrade): Do you wash the car before or after your ride?

 

[thecheapseats]

Firmware images do get validated when being flashed. Both the actual model as well as a CRC32 integrity check to ensure the image isn't corrupted. Anyone who managed to flash an AX86U firmware on an AC86U must have done something really unusual to flash it, like using TFTP...

of course they get validated - I agree, it doesn't make sense... chcksum and mod#... how the heck did that happen??... in a private message, if you have any cli suggestions for him you'd rather not post up here 'in-the-wild' if he goes ttl serial, it might help him out of brickland... I got nuthin' even if I was sitting at a ttl serial prompt other than forcing a tftp that bypasses the nic hardware...

 

[RMerlin]

of course they get validated - I agree, it doesn't make sense... chcksum and mod#... how the heck did that happen??... in a private message, if you have any cli suggestions for him you'd rather not post up here 'in-the-wild' if he goes ttl serial, it might help him out of brickland... I got nuthin' even if I was sitting at a ttl serial prompt other than forcing a tftp that bypasses the nic hardware...

TFTP in recovery mode is the only way to do it. If it fails to enter recovery mode, then it means either he has broken hardware, or somehow he overwrote his bootloader in which case there is nothing he can do.

 

[thecheapseats]

TFTP in recovery mode is the only way to do it. If it fails to enter recovery mode, then it means either he has broken hardware, or somehow he overwrote his bootloader in which case there is nothing he can do.

hmm - no way to revive the cfe btldr? - ok - I know that's in the weeds - sorry... and before the re-read I thought he'd bricked a new ax86 - not an old ac86 with high-time on it... oops on my part - had it backwards... but still ; ) ...

 

[Sky1111]

Folks I noticed something unusual about Beta5: Internet speed on wireless clients seem to be significantly lower (than the wired ones)
So on wired ones and the router itself I do get 930-1000 Mbps out of gigabit internet (when family is not active), so that one is perfect.
On a wireless AX ThinkPad that is next to mesh node and is connected at 2402/2402 Mbps, I only see ~520-630 Mbps, using the same Fast.com. That is Ryzen Pro 4750u fast laptop with pretty fast NVMe drive, and with Beta3 on the same connection, I did clock Internet close wired speed!
What is more interesting, is that when I do a local large file copy from my home server to the laptop over wifi, I do saturate gigabit network that the server is connected to the router, so that 2402 Mbps connection is very real

Fast AC clients that connect at solid 866Mbps, are maxing out at 350.

With 2.4Ghz and 300Mbps connection, it maxes out at 150-180 (tested on 2 fast ThinkPads).

And this experience is consistent - no wireless client in the household get proper Internet speed. Of course wifi capabilities, etc vary accross the devices, but I start noticing the pattern.

So why is such a discrepancy between wifi connections, real data transfer rates for wifi clients (from wired LAN hosts), and wireless Internet speeds? I do not expect wifi clients to be as fast as wired, but when you see that you internet connection speed is like half of data transfer capability, that seems strange.
Am I the only one seeing this?

 

[joegreat]

Fast AC clients that connect at solid 866Mbps, are maxing out at 350.

With 2.4Ghz and 300Mbps connection, it maxes out at 150-180 (tested on 2 fast ThinkPads).

A throughput around the half of the link speed is also my experience - but already with past firmware and actual ones the same...
So, it does not seem to be an issue in the current beta version, right?

 

[WET]

Posting this to find out if anyone else is having the same issue with 386.1 releases that I am.

This has been / still is happening with all Merlin 386.1 alpha's & beta's. I have one client, a W10 PC with an Intel AC7260 wireless adapter, that constantly looses it's 5Ghz wireless connection. It takes a while for the drop to happen after initial PC boot up but after that drops gets closer and closer together. No other clients on RT-AC86U or AIMesh node RC-AC68U drop.

This client is very close to the 86U, 6ft. Once dropped I have to reset the client in the PC via running network diagnostics or manually disable / enable wireless client. I have forgotten wireless network connection several times within PC. I have tried disabling both beamforming's in router, it does not make a difference. This client has always been rock solid prior to this release. I believe there was a new driver version for the 86U that came out with the 386.1 release. I've been hoping that newer releases will fix.

I did do a factory reset after flashing to 386.1 alpha as instructed. The drops have gotten so bad I am now using an Ethernet connection. I read the forum often enough to know no that is not an issue that is being encountered by many.

Any help / suggestions will be appreciated.

For what it is worth, I have figured out what the problem was. I had a bad Intel AC7260 Client in the PC, When I swapped out the client mini PCIe no more disconnects.

 

[maxbraketorque]

A throughput around the half of the link speed is also my experience - but already with past firmware and actual ones the same...
So, it does not seem to be an issue in the current beta version, right?

A quick test of my MacbookPro with 3x3 AC shows its internet download speed maxing out at about half the link speed, i.e., maxing out at about 600 mbps for a link speed of 1300 mbps with my AC86U. This is on a 1 gbps connection that will reach 930 mbps on my wired computer. Out of curiosity, I tried with pwr wait disabled and got the same result. In both cases, CPU core 1 on the router goes to about 80% usage while core 2 initially spikes up to around 60% and then drops and hovers at about 15% usage.

 

[maxbraketorque]

@RMerlin, a few official firmwares higher than 41700 have been released. Do you know what changes these have, and are you considering incorporating the next unified GPL release? I know this can be a never-ending game, but still want to ask since this has been such a long beta period.

 

[GHammer]

Does that allow you to limit reservations to a specific interface, or are you forcing that client to be assigned that IP no matter what interface they're connected to?

Do a wee bit of investigating.

 

[RMerlin]

@RMerlin, a few official firmwares higher than 41700 have been released. Do you know what changes these have, and are you considering incorporating the next unified GPL release? I know this can be a never-ending game, but still want to ask since this has been such a long beta period.

I currently have no plans to merge any newer GPL for 386.1. 41700 is good enough to be used for a formal release.

 

[heysoundude]

I currently have no plans to merge any newer GPL for 386.1. 41700 is good enough to be used for a formal release.

I'm reading this like it'll be a "this weekend" thing...

 

[mvadu]

I don't think it is directly from @RMerlin branch, but AiMesh 2.0 and Smart Connect seems to have some problems in routing clients to proper nodes/bands. I have ac-3100 as main and a ac-86u as mesh node. I had Smart Connect enabled since beginning and with old 384 code all my devices were happy.

Since going to beta 5, my kids Chromebook started acting crazy, even though its next to ac3100 it will try to connect to upstairs 86u and says "medium signal strength" and it can't even access internet. Nothing in the logs that I can correlate.

Turning off mesh node gets everything back online. Today I tried turning off the smart connect, and so far things are okay even with mesh node ON.

 

[L&LD]

Update: Smart Connect, still not as smart as it should be in 2021.