Beta Asuswrt-Merlin 386.2 Beta is now available

[cristobal07]

maybe it's the same issue goodwin_c, Chalnoth and AntonK mentioned,
that I'm facing too.

rt-ac88u > upgraded from 386.2 alpha 2 to beta 1 > factory reset > reconfigured

gui shown internet status connected, but no connection on clients end.
disconnect and reconnect wan connection from gui did the trick.

having DoT enabled as well.

Does the system log report anything regarding the WAN status when it drops?

saw this entry on initial connection :

WAN_Connection: Fail to connect with some issues.

and 2 repeated entries as below :

kernel: ADDRCONF(NETDEV_UP): ifb0: link is not ready
kernel: ADDRCONF(NETDEV_UP): ifb1: link is not ready
kernel: ADDRCONF(NETDEV_UP): fwd0: link is not ready
kernel: ADDRCONF(NETDEV_UP): agg: link is not ready
kernel: ADDRCONF(NETDEV_UP): dpsta: link is not ready

thank you.

 

[skeal]

can someone suggest right values for Cake overhead FTTH GPON , Modem in bridge mode, PPPOE mode in ASUS AX88U ?

I would try Ethernet/Vlan it's what I use and I have FTTH

 

[elorimer]

A very, very small cake thing coming from .1_2 with the cli version. I am provisioned at 230/35, and had the cli version set at 200/30

I uninstalled the cli version, installed .2 Beta. The default Cake settings showing were 30/200. I didn't notice at first since the numbers were familiar, so the speedtests at 28/35 stumped me a bit. I don't know if the 30/200 was a carryover from the cli version? But reversed? When I corrected them to 200/30 everything was fine. (also: everything else looks fine).

86U+68U (aimesh node)

 

[visortgw]

A very, very small cake thing coming from .1_2 with the cli version. I am provisioned at 230/35, and had the cli version set at 200/30

I uninstalled the cli version, installed .2 Beta. The default Cake settings showing were 30/200. I didn't notice at first since the numbers were familiar, so the speedtests at 28/35 stumped me a bit. I don't know if the 30/200 was a carryover from the cli version? But reversed? When I corrected them to 200/30 everything was fine. (also: everything else looks fine).

86U+68U (aimesh node)

The down/up speeds for QOS have always been intuitively backwards in the router GUI.

 

[prosperot]

maybe it's the same issue goodwin_c, Chalnoth and AntonK mentioned,
that I'm facing too.

rt-ac88u > upgraded from 386.2 alpha 2 to beta 1 > factory reset > reconfigured

gui shown internet status connected, but no connection on clients end.
disconnect and reconnect wan connection from gui did the trick.

having DoT enabled as well.



saw this entry on initial connection :

WAN_Connection: Fail to connect with some issues.

and 2 repeated entries as below :

kernel: ADDRCONF(NETDEV_UP): ifb0: link is not ready
kernel: ADDRCONF(NETDEV_UP): ifb1: link is not ready
kernel: ADDRCONF(NETDEV_UP): fwd0: link is not ready
kernel: ADDRCONF(NETDEV_UP): agg: link is not ready
kernel: ADDRCONF(NETDEV_UP): dpsta: link is not ready

thank you.

Yes , it looks like the same problem. Resetting the ax88 did not work. Will try resetting the WAN later on.

 

[AntonK]

Does the system log report anything regarding the WAN status when it drops?

How are you testing the Internet connection? Check by trying to ping an IP address like 8.8.8.8 to see if it's a DNS issue or really a connectivity issue.

The only WAN-related change in 386_42095 that I am aware of is now Asus configures static routes to the WAN DNS servers.

Hi @RMerlin . I regret not having captured my logs. I'll remember this for the future so my (our) experiences are of help to you. But, I strongly believe that it was a DNS issue. Everything showed as being connected to the internet (e.g. pc, iphone, Echo), and while the Echo continued to function, browsing with any other device did not work. Any attempt to browse would just not go anywhere, but no errors or timeouts that I experienced either. If I'd been thinking, I would've tried a straight IP test to somewhere so as to get around DNS. I bet my eye-teeth an IP in the address bar would've worked. I've had DNS resolution failures in the somewhat distant past. This felt exactly like those.

 

[elorimer]

The down/up speeds for QOS have always been intuitively backwards in the router GUI.

Yes, I follow that. What I meant was 200mbps was showing initially as my upload speed, and 30mbps was showing as my download speed. If that is the a priori assumption in the firmware, I suggest that is backwards, because most people are assymetrical the other way. But if it was populating from something that persisted after I uninstalled the cake cli, then it is also backwards.

 

[ttgapers]

A very, very small cake thing coming from .1_2 with the cli version. I am provisioned at 230/35, and had the cli version set at 200/30

I uninstalled the cli version, installed .2 Beta. The default Cake settings showing were 30/200. I didn't notice at first since the numbers were familiar, so the speedtests at 28/35 stumped me a bit. I don't know if the 30/200 was a carryover from the cli version? But reversed? When I corrected them to 200/30 everything was fine. (also: everything else looks fine).

86U+68U (aimesh node)

Shouldn't use both concurrently. If you use the mainline in the firmware, you can set the script to "Stop" (which is what I have done for now).

 

[kabukimurder]

Hey @RMerlin i have a question.
When i run my X86U with "Traffic Analyzer - Statistic" enabled my CPU clearly starts to check the traffic with a 70% CPU load as a result.
It makes sense that HW acceleration would be disabled with this feature on, but on the Tools-page it still says: Runner: Enabled - Flow Cache: Enabled

The original firmware throws out a log post saying that FC is disabled when traffic starts going over the CPU as a feature is enabled, this however seems suppresed in the Merlin versions.

My question is: Is the Tools-page status for HW-acceleration not working correctly. Or can both of these HW Acc features really still be enabled?

 

[elorimer]

Shouldn't use both concurrently. If you use the mainline in the firmware, you can set the script to "Stop" (which is what I have done for now).

I uninstalled the cli version, installed .2 Beta.

I guess I'm not very good at communicating this. I'm not using both. I uninstalled the addon before installing the beta firmware. What I am positing is that the beta firmware, when cake as a QOS is first selected, might be picking up old settings from the addon, but incorrectly.

 

[RMerlin]

Do you know what fw 386.42095 equivalent to AX86U & AC86U's?

42095 is 42095. It's possible Asus simply hasn't released a firmware for these specific models based on that exact code, so there are now equivalent atm.

 

[visortgw]

42095 is 42095. It's possible Asus simply hasn't released a firmware for these specific models based on that exact code, so there are now equivalent atm.

Asus has now released 42095 for the RT-AX58U/RT-AX3000 -- it appears to be a work in progress.

 

[RMerlin]

My original WAN/LAN DNS was served my two Bind9 servers that I run on my LAN (they aren't routed through the VPN). I've ran it that way for years, and have only had the issue on the last alpha that was released, and on this beta.

Asus now sets up static routes for WAN DNS servers, for some unknown reason. It's possible that these routes didn't work as expected with your previous DNS servers.

Hi @RMerlin could you please add the GPL version info on the Sysinfo page in your firmware?

No, because that version isn't always cut dry. Some releases may have a mixture of multiple versions. That information is available in the changelog, and I expect users to read it before they flash a new firmware.

Small suggestion. When alpha and betas are released include a template for reporting problems. Each user who wants to report a "problem" has to use the template.

We recently tried that on Github in an attempt to make the Issue tracker semi-useful. 75% of reporters simply remove the template and post whatever they want...

 

[ttgapers]

I guess I'm not very good at communicating this. I'm not using both. I uninstalled the addon before installing the beta firmware. What I am positing is that the beta firmware, when cake as a QOS is first selected, might be picking up old settings from the addon, but incorrectly.

Gotcha and thanks for clarifying. I am not experiencing this though. Curious if others are.

 

[RMerlin]

My question is: Is the Tools-page status for HW-acceleration not working correctly. Or can both of these HW Acc features really still be enabled?

The report on the Sysinfo page should actually be more accurate as of the last two or three releases since I now directly ask the driver to report their state. So if it says enabled, then it must really be.

 

[maxbraketorque]

Updated one set of my two AC86U main/AP combos without issue. Nothing unusual in the router logs, and everything appears to be functioning normally. Fairly minimal setup. Using built-in DoT, two continuously running OVPN clients on the router, some port forwarding, some static IPs and manual DHCP IP addresses on the LAN, using DDNS and Let's Encrypt, DNSFiltering enabled, Traffic Analyzer enabled, Parental Controls enabled.

No reductions in CPU temperature over the last few 386.2 distributions. Temperature remains about 4-5C higher than 384.19. But temperature is only ~80C, so no prob.

 

[elorimer]

Gotcha and thanks for clarifying. I am not experiencing this though. Curious if others are.

I poked at this some more. I disabled qos in the firmware. Then I deleted /etc/cake-qos.conf. Rebooted. Installed the CLI version, set the bandwidth. Then, stopped it. Rebooted.

Then I re-enabled the firmware version, and it picked up what clearly were upload/download and the other parameters from the original configuration and recreated /etc/cake-qos.conf.

So I'm going to mark this as of no further importance. The next step would have been to roll back restore backups and roll forward.

 

[jpthsd]

One of the "hidden" advantages of adopting Merlinware is that you often get the latest Asus GPL code even though that particular version may not be released yet for your model [and sometimes the version numbers for different models never match on the Asus side].

This particular beta will [I'm guessing] be based on the Asus GPL released for the RT-AX88U ...
View attachment 31836
Another accolade for the Merlin maestro - we get to test a beta within a week of the Asus GPL release ... for a whole range of models .

Thank you very much for the information. I am glad to know about this!

I hope with 386.2 will resolve the AC86U (as AiMesh node) buffering since I continued getting while on 386.1 and even 386.1_2 (thus switched back to AsusWRT for both AX86U & AC86U which have been stabilized 19 days without any hiccup)!

Thanks again!

 

[Darcy]

Now that the AX11000 is supported is it worth it without the ROG features?

 

[kabukimurder]

The report on the Sysinfo page should actually be more accurate as of the last two or three releases since I now directly ask the driver to report their state. So if it says enabled, then it must really be.

I rebooted the router and confirmed that Tools -> HW acceleration was Enabled still. And that the CPU is working pretty hard when i download full clip.
This is what the startup logs say:

May 5 07:05:08 kernel: ^[[0;36;44mTotal # of labels = 68^[[0m
May 5 07:05:08 kernel: ^[[0;36;44mInitialized fcache state^[[0m
May 5 07:05:08 kernel: Pkt HW acceleration is disabled/unavailable.
May 5 07:05:08 kernel: ^[[0;36;44mBroadcom Packet Flow Cache Char Driver v4.0 Registered<302>^[[0m
May 5 07:05:08 kernel: fc_timer_task created successfully
May 5 07:05:08 kernel: Pkt HW acceleration is disabled/unavailable.
May 5 07:05:08 kernel: Created Proc FS /procfs/fcache
May 5 07:05:08 kernel: ^[[0;36;44mBroadcom Packet Flow Cache registered with netdev chain^[[0m
May 5 07:05:08 kernel: ^[[0;36;44mBroadcom Packet Flow Cache learning via BLOG enabled.^[[0m
May 5 07:05:08 kernel: ^[[0;35m[FHW] pktDbgLvl[0xffffffbffc481c08]=0^[[0m
May 5 07:05:08 kernel: ^[[0;34m[FHW] fhw_construct: ^[[0m
May 5 07:05:08 kernel: ^[[0;36;44mInitialized Fcache HW accelerator layer state^[[0m
May 5 07:05:08 kernel: flwStatsThread created. numFlwsPerPoll 546 maxFlwIdx 16383
May 5 07:05:08 kernel: ^[[0;36;44mFcache Pathstats created^[[0m
May 5 07:05:08 kernel: ^[[0;36;44mConstructed Broadcom Packet Flow Cache v4.0^[[0m
(**some unrelated log rows** )
May 5 07:05:08 kernel: Retrieve num_fhw_path=64
May 5 07:05:08 kernel: ^[[0;36;44mPathstats allocated 2048 bytes^[[0m
May 5 07:05:08 kernel: ^[[0;36;44mBroadcom Packet Flow Cache HW acceleration enabled.^[[0m
May 5 07:05:08 kernel: Enabled Runner binding to Flow Cache
May 5 07:05:08 kernel: Initialized Runner Protocol Layer (800)
May 5 07:05:08 kernel: Broadcom Runner Blog Driver Char Driver v0.1 Registered <309>

Could it be that NAT acceleration is working but the CPU is still sweating to populate Traffic Analyzer Statistics? Is the statistics even supposed to be compatible with Hardware accelerated NAT?