[Fork] Asuswrt-Merlin 374.43 LTS releases (Archive)

[dave14305]

We all have choices in life. For those of you who run John's fork on a RT-AC68U, what is the primary reason? Since the AC68U is still supported by Merlin on the 384 branch, it's not the same scenario as running the fork on an N66U or old AC66U where you have no viable options.

I enjoy the stability and amount of refinement that has gone into the fork. Not having to chase the latest ASUS GPL allows energy to be spent on fixing many smaller issues without worrying about later merges. However, there are some downsides like having to disable HW acceleration to enable QoS. Yet my ISP bandwidth is still acceptable without it, and John's QoS is great once you learn how to configure it.

Running Merlin gives the latest features, but it feels like the ground is always moving under your feet with GPLs, alphas, betas, final releases. The quality is high, but the amount of churn is also high. Yet to be a valuable contributor to the discussions, I feel it's better to run the majority firmware.

I also bemoan the fact that John's repo is not searchable on Github since his fork has fewer stars than Merlin's original repo. I like to study the code and not being able to search it is a barrier to my own learning. I end up searching Merlin's 380 branch then looking in the comparable location in John's fork.

Looking forward to @john9527's return to the forums one day soon, hopefully.

 

[Grisu]

For those of you who run John's fork on a RT-AC68U, what is the primary reason?

some will like old wireless drivers with much more power and coverage.
Others maybe seeing the end of 68U support from asus and so merlin and hopefully better change now than later.

 

[Lislet]

Looking forward to @john9527's return to the forums one day soon, hopefully.

Long time John's fork user. @john9527, thank you for all your efforts.

@dave14305 any idea what's going on? I hope it isn't anything severe? I heard John had some hard time in his family recently. Hope for the best.

 

[ColinTaylor]

Long time John's fork user. @john9527 , thank you for all your efforts.

@dave14305 any idea what's going on? I hope it isn't anything severe? I heard John had some hard time in his family recently. Hope for the best.

I'm sure we're all in agreement in offering our gratitude and best wishes to John. Beyond that, he posted here that he was going to be busy for a while so I think we should be patient and respect his privacy. I take some comfort in seeing that his account was logged into the forum 2 weeks ago even if he didn't post anything.

 

[Gar]

Is there a local dns cache setting in 374.43, I don't see it? Big topic for the 384.11/12 fw.

 

[dave14305]

Is there a local dns cache setting in 374.43, I don't see it? Big topic for the 384.11/12 fw.

No, it only came to Merlin in 384.10. It hasn’t been backported to Johns fork. Not sure it’s necessary.

 

[Gar]

No, it only came to Merlin in 384.10. It hasn’t been backported to Johns fork. Not sure it’s necessary.

John has another way of handling it, or you're not convinced it's necessary at all?

 

[dave14305]

John has another way of handling it, or you're not convinced it's necessary at all?

The router will use dnsmasq as long as dnsmasq started successfully. Otherwise it reverts to the WAN DNS. But if dnsmasq didn’t start, it means you have problems.

Personally I think the new option has caused a lot of confusion, even though it has technical merit. I’m not sure John wants confusion in his LTS fork.

 

[chigo58]

Hello all - forgive me for asking if these questions have already been answered. I've loaded the latest E variant on my AC66U - no problems in wireless router mode - just a few questions:

- Is it possible to assign the WAN port as a LAN switch port? (haven't managed to find this option)
- WAN is disabled, is it possible to get the router to sync with an NTP router if WAN is disabled?

Thank you

 

[ColinTaylor]

- Is it possible to assign the WAN port as a LAN switch port? (haven't managed to find this option)

Switch to Access Point mode.

 

[dave14305]

As I get comfy again on John’s fork, I’ve been trying to come up with a suitable traditional QoS ruleset. In studying the QoS posts in the thread and elsewhere on the internet, I came up with the rules below. Not a lot of requirements other than TV streaming is higher than downloads, but lower than regular (small) web traffic. There’s not a lot of differentiation anymore in traffic now that HTTPS is nearly everywhere. Does anybody use any other interesting rules for QoS? The only non 80/443 traffic I see is usually 5223/tcp for our Apple devices, so that’s covered already.

Hopefully this image resolution doesn’t suck:

 

[Gar]

Time for a update to his firmware, hope he gets back soon.

 

[ColinTaylor]

Time for a update to his firmware, hope he gets back soon.

Why? What problems have you found in the current release?

 

[ColinTaylor]

Hopefully this image resolution doesn’t suck:

FWIW here is my current ruleset. Like you my main concern was trying to figure out how to prioritise video streaming sites. Ultimately I decided that it's pretty much impossible to distinguish streaming from something like a Steam download. The problem with something like Netflix is that it holds open the HTTPS connection for a long time so most of it's traffic would be classified as a "bulk download" (see third image). Even though this traffic is "bursty" giving it anything lower than a medium priority means that a) almost everything else will interfere with it, and b) the packet scheduling is really bad. In the end the best compromise for me was to set the HTTPS traffic to High/Medium priority in combination with a default of Low.

 

[Swistheater]

If I wanted to bring a 56U merlin 384.6 to john fork, what would be the best instructions aside from the factory resets that maybe involved in the process?

 

[ColinTaylor]

If I wanted to bring a 56U merlin 384.6 to john fork, what would be the best instructions aside from the factory resets that maybe involved in the process?

As per the @UpgradeMatrix.txt document, you would need to use the Restoration Tool (recovery mode) to upload the firmware followed by a factory reset and manual setup.

 

[Swistheater]

my next question would be does freshjr work with johns fork?

 

[ColinTaylor]

my next question would be does freshjr work with johns fork?

No. FreshJR is a mod for Adaptive QoS. John's firmware doesn't have Adaptive QoS, only Traditional QoS.

 

[dave14305]

FWIW here is my current ruleset. Like you my main concern was trying to figure out how to prioritise video streaming sites. Ultimately I decided that it's pretty much impossible to distinguish streaming from something like a Steam download. The problem with something like Netflix is that it holds open the HTTPS connection for a long time so most of it's traffic would be classified as a "bulk download" (see third image). Even though this traffic is "bursty" giving it anything lower than a medium priority means that a) almost everything else will interfere with it, and b) the packet scheduling is really bad. In the end the best compromise for me was to set the HTTPS traffic to High/Medium priority in combination with a default of Low.

Thanks for that. I am looking into /proc/net/ip_conntrack to observe the byte counts of connections and checking the hit-rate of the QOSO rules running "watch -n 3 iptables -t mangle -L QOSO -v -n" in a PuTTY session. But that gets hard once the connection is properly marked.

For me, since I work from home most of the time, my priority is that the family streaming Amazon Prime movies on the TV, or YouTube videos on other devices would not choke my work laptop that may be using Skype for Business (audio/screensharing) but mostly while connected to corp VPN. So I've added a rule for my laptop MAC to be in the High category.

Is there an "easy" way to gauge the effectiveness of the rules? How did you make that fancy graph? nudge nudge

I need to also noodle on the bandwidth priority min/max percentages once I decide what's going to be in each priority level. I'm mostly default there, but was thinking most maximums could be 100% to allow them full bandwidth if no competition (borrowed from FreshJR's posts/script). Is that flawed thinking with Traditional QoS?

Thanks again Colin.

 

[ColinTaylor]

Is there an "easy" way to gauge the effectiveness of the rules? How did you make that fancy graph? nudge nudge

I don't know of an easy way. Like you I started by looking at the packet counts but that is onerous which is why I created those graphs. Like a lot of my code it was just a quick and dirty botch that sends the statistics to a Linux server which in turn produces the graphs. Testing the effectiveness was really a challenge, particularly because my bandwidth is so large. The only way I could do it was by limiting the total download bandwidth to 10Mbps and then starting long running downloads on two separate PCs that had been assigned different priorities. Here's an image from one of those tests:

I need to also noodle on the bandwidth priority min/max percentages once I decide what's going to be in each priority level. I'm mostly default there, but was thinking most maximums could be 100% to allow them full bandwidth if no competition (borrowed from FreshJR's posts/script). Is that flawed thinking with Traditional QoS?

Apart from the Lowest priority I ran for a long time with the other maximums set at 100%. In theory this should be fine given the reserved minimums, but my "feeling" (after a year or so) was that very large downloads of unclassified traffic ("Default=Low" in my case) was still having a negative impact on the higher priority traffic so I capped that at 80%.