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

[badboyme]

@john9527


Hi John,
I would like to submit the following problem to you:
I live in Italy and I have a Vodafone italian ISP's FTTH internet connection;
as many italian ISPs, Vodafone provides to me a GPON ONT (Huawei) and a VDSL Modem/Router VSR (Vodafone Station Revolution - also made by Huawei) in leasing.
The VSR is connected to the ONT in the following manner: VSR (WAN port) <----> ONT (LAN port) and ONT (GPON port) <---> Fiber cable that enter to my home.
I would like to use my ASUS RT-AC68u router in place of the buggy VSR Vodafone router, but there is one problem: beetween VSR and ONT there is a PPP session with a Vodafone PPP Host-Unique-tag that creates a non standard method to force Vodafone's users to use the VSR only to establish a connection (or, in others words, you normally can't use a router different from the VSR).
Reading around the italian forums, I seen that Vodafone uses a unknown username and the HEX of VSR S/N as password to establish the PPP internet session.
My question is: is it possible to add a code/script (with a mini-guide) to insert the 2 mentioned parameters in the PPPD process/daemon of your modded Asus firmware to use ASUS Router in place of the VSR Vodafone Router?

Here I found a link with some useful code (Sorry, but I am not a programmer): LINK

Thank you in advance!!
Best Regards from Italy!

 

[john9527]

The Update is unstable than beta 1, cont refresh the connection details

Can't recreate any problem on my end. If you are using Firefox as your browser, you may need hit the refresh button twice to have it display due to a bug in FF. IE/Edge and Chrome don't have this problem.

 

[john9527]

I live in Italy and I have a Vodafone italian ISP's FTTH internet connection;
as many italian ISPs, Vodafone provides to me a GPON ONT (Huawei) and a VDSL Modem/Router VSR (Vodafone Station Revolution - also made by Huawei) in leasing.
The VSR is connected to the ONT in the following manner: VSR (WAN port) <----> ONT (LAN port) and ONT (GPON port) <---> Fiber cable that enter to my home.
I would like to use my ASUS RT-AC68u router in place of the buggy VSR Vodafone router, but there is one problem: beetween VSR and ONT there is a PPP session with a Vodafone PPP Host-Unique-tag that creates a non standard method to force Vodafone's users to use the VSR only to establish a connection (or, in others words, you normally can't use a router different from the VSR).

This sounds like a twist on one of the Movistar IPTV profiles. I had to bypass supporting Movistar as it's too big a change (over 1000 lines of code all over the place). I think your best hope here is to try and submit a support request directly to ASUS, then Merlin can potentially pick it up in one of his merges for his builds.

 

[slobodan]

Perhaps I was not very explicit, but dnscrypt dies unexpectedly and stays dead. I did not have this problem with dnscrypt from entware-ng.

 

[john9527]

Perhaps I was not very explicit, but dnscrypt dies unexpectedly and stays dead. I did not have this problem with dnscrypt from entware-ng.

I've never seen it 'die'....how long before you see it have a problem? also post up the system log somewhere for me to take a look at.

EDIT: Also please change the log level for dnscrypt to 'Info' if you haven't already done so.
EDIT2: Another question, were you doing any other configuration changes on the router prior to the problem? It may be that I missed handling a restart correctly on some operation.

 

[slobodan]

I've never seen it 'die'....how long before you see it have a problem? also post up the system log somewhere for me to take a look at.

EDIT: Also please change the log level for dnscrypt to 'Info' if you haven't already done so.
EDIT2: Another question, were you doing any other configuration changes on the router prior to the problem? It may be that I missed handling a restart correctly on some operation.

I will gather the logs later.

dnscrypt usually died after some minutes after rebooting the router. I had then an entware service killing dnscrypt and then restarting it, this ensured that it worked for something less than a day.

 

[slobodan]

I've never seen it 'die'....how long before you see it have a problem? also post up the system log somewhere for me to take a look at.

EDIT: Also please change the log level for dnscrypt to 'Info' if you haven't already done so.
EDIT2: Another question, were you doing any other configuration changes on the router prior to the problem? It may be that I missed handling a restart correctly on some operation.

Tried to send syslog.log by PM, got a message that I got blocked.

 

[fink.nottle]

@fink.nottle What version of the firmware are you using? Traditional QOS didn't work properly until beta version 22B2. Try loading the latest beta then it should work.

I'm on the latest stable (20E9). That could explain it. I'll try the beta soon. I'm wondering if I should use a modern linux router, and use asus just as an access point. I don't know if that will be better. Do you have better luck with the latest beta w.r.t. bufferbloat ?

 

[ColinTaylor]

I'm on the latest stable (20E9). That could explain it. I'll try the beta soon. I'm wondering if I should use a modern linux router, and use asus just as an access point. I don't know if that will be better. Do you have better luck with the latest beta w.r.t. bufferbloat ?

No problem with buffer bloat (not that I had a problem before anyway)

 

[john9527]

Tried to send syslog.log by PM, got a message that I got blocked.

post it to a file sharing site and PM me the link.

DNSCrypt will normally start at least twice during a boot cycle (once to operate in a mode such that it can work with the system time not being set, and then again when the system time is set correctly).

It's also tied into the state of dnsmasq. It sounds as if you are savvy enough to be doing custom scripting. If you are doing anything outside of the standard Merlin custom script support that effects dnsmasq, you may need to restart dnscrypt manually via 'service restart_dnscrypt'.

 

[slobodan]

post it to a file sharing site and PM me the link.

DNSCrypt will normally start at least twice during a boot cycle (once to operate in a mode such that it can work with the system time not being set, and then again when the system time is set correctly).

It's also tied into the state of dnsmasq. It sounds as if you are savvy enough to be doing custom scripting. If you are doing anything outside of the standard Merlin custom script support that effects dnsmasq, you may need to restart dnscrypt manually via 'service restart_dnscrypt'.

I have sent you a PM and reposted it in a conversation.

 

[Futuristic]

Hey quick question for you guys. I'm running 20e9 on an N66U over here and all is well but I've started seeing this in the logs in the last couple days and it's completely flooding the log up every minute or so and I'm not sure what it means. Pretty basic user with a large amount of devices on the network. Just looking for some help on it and how I can make it stop. Not running any IPv6, vpn or anything special. Thanks guys.

Dec 17 02:00:15 miniupnpd[580]: Failed to convert hostname 'fe80::290:a9ff:fed9:cac6' to ip address

 

[john9527]

Hey quick question for you guys. I'm running 20e9 on an N66U over here and all is well but I've started seeing this in the logs in the last couple days and it's completely flooding the log up every minute or so and I'm not sure what it means. Pretty basic user with a large amount of devices on the network. Just looking for some help on it and how I can make it stop. Not running any IPv6, vpn or anything special. Thanks guys.

Dec 17 02:00:15 miniupnpd[580]: Failed to convert hostname 'fe80::290:a9ff:fed9:cac6' to ip address

That is definitely a new one. To start, that MAC/link local IPv6 address is owned by Western Digital. Have you added or updated the firmware on a WD NAS or CloudDrive recently?

 

[fink.nottle]

No problem with buffer bloat (not that I had a problem before anyway)

That looks really good. I haven't had the time to try the john's fork's beta yet, but I tried the latest mainline merlin, and it doesn't do anything either. Interestingly shibby-tomato manages to solve it for me, to an extent. For ethernet, tomato does a very good job of handling this. I have a 100/10 Mbps internet connection, which is overprovisioned by 20% (so 120/12ish in reality). In tomato, after enabling QoS, and setting the limits as 120/12, I get speeds close to 100/10 on ethernet and simultaneous pings under 30ms, which is great. The only downside is that on wifi, the download drops to something like 75 Mbps after enabling QoS. @john9527 , any idea what's going on ? What does tomato do differently ? I remember reading a while ago that fq_codel/cake cannot be integrated too well into these asus routers due to old kernel, drivers, blobs etc.

 

[Marctraider]

Ive been using this fork to get better reachability especially with 5Ghz in some areas in my house and it works great. However a few days ago i stopped having any connectivity from specific areas which led me to believe something reset the transmit power.

Is there a known issue with this? Perhaps it rebooted due to a power loss of course which resets the transmit power settings. I guess I should make it apply every time it reboots?

 

[john9527]

That looks really good. I haven't had the time to try the john's fork's beta yet, but I tried the latest mainline merlin, and it doesn't do anything either. Interestingly shibby-tomato manages to solve it for me, to an extent. For ethernet, tomato does a very good job of handling this. I have a 100/10 Mbps internet connection, which is overprovisioned by 20% (so 120/12ish in reality). In tomato, after enabling QoS, and setting the limits as 120/12, I get speeds close to 100/10 on ethernet and simultaneous pings under 30ms, which is great. The only downside is that on wifi, the download drops to something like 75 Mbps after enabling QoS. @john9527 , any idea what's going on ? What does tomato do differently ? I remember reading a while ago that fq_codel/cake cannot be integrated too well into these asus routers due to old kernel, drivers, blobs etc.

Lot's of questions in there
- re:wifi.....a couple of possibilitie...either a

• driver difference, my fork and Merlin use the ASUS OEM drivers (but different versions), I think tomato uses an open driver.
• you didn't mention speeds without qos or what type of wifi adapter in your client, but 75Mbps is about right for a simple 1x1 N adapter

- re QoS implementations....they all are different. I use a method that can be common to MIPS as well as ARM routers. Shibby uses different implementations based on the type of router. The original ASUS implementation had some limitations when dealing with downloads.

- re:queuing disciplines...we were able to backport support for codel/fq_codel to the ARM based routers (not possible for the older kernel used on the MIPS routers). It was backported from someone who worked it out for tomato. Not aware anyone has gone to any of the newer disciplines or if that is going to be possible.

 

[john9527]

Ive been using this fork to get better reachability especially with 5Ghz in some areas in my house and it works great. However a few days ago i stopped having any connectivity from specific areas which led me to believe something reset the transmit power.

Is there a known issue with this? Perhaps it rebooted due to a power loss of course which resets the transmit power settings. I guess I should make it apply every time it reboots?

Nothing there that I am aware of. In fact, it's recommended that you reboot if you change the transmit power in the gui and don't see any change to make sure it takes.

Hate to say it, but if it occurred after a power fail either the router or it's power supply may have been damaged (are you using any type of surge protection?). Also, I seem to remember a couple of sporadic reports of nvram corruption after a power fail (across all firmware). If you have a saved config, you might try factory reset/restoring it to make sure).

 

[Futuristic]

That is definitely a new one. To start, that MAC/link local IPv6 address is owned by Western Digital. Have you added or updated the firmware on a WD NAS or CloudDrive recently?

Thanks so much, John!! We do in fact have a western digital cloud drive that looks like it updated its firmware 2 nights ago! Ha! Nice detective work there!! I've rebooted it and it appears the message has disappeared.

Thank you for your help and awesome firmware!! Really appreciate it!

 

[john9527]

Thanks so much, John!! We do in fact have a western digital cloud drive that looks like it updated its firmware 2 nights ago! Ha! Nice detective work there!! I've rebooted it and it appears the message has disappeared.

Thank you for your help and awesome firmware!! Really appreciate it!

Whew.....glad to hear it! If it wasn't something simple like that, then it would be ugly to figure out.

 

[fink.nottle]

Lot's of questions in there
- re:wifi.....a couple of possibilitie...either a

• driver difference, my fork and Merlin use the ASUS OEM drivers (but different versions), I think tomato uses an open driver.
• you didn't mention speeds without qos or what type of wifi adapter in your client, but 75Mbps is about right for a simple 1x1 N adapter

- re QoS implementations....they all are different. I use a method that can be common to MIPS as well as ARM routers. Shibby uses different implementations based on the type of router. The original ASUS implementation had some limitations when dealing with downloads.

- re:queuing disciplines...we were able to backport support for codel/fq_codel to the ARM based routers (not possible for the older kernel used on the MIPS routers). It was backported from someone who worked it out for tomato. Not aware anyone has gone to any of the newer disciplines or if that is going to be possible.

My wifi card is Centrino Ultimate-N 6300, which is a 3x3 card. It's a pretty shirtty card (at least on linux), which I can't replace easily due to bios whitelist, but it does manage to hit ~100 Mbps on your fork as well as tomato (without QoS). It doesn't do that on mainline Merlin anymore (on linux. On windows it can). However, what's interesting with tomato is after enabling QoS, the wifi speeds drop from 100ish to 75, whereas ethernet drops from 120ish to 100ish. I thought all the QoS happens at the WAN side. i.e. I wasn't expecting wifi to be affected separately. Guess that's not how it works. The other interesting thing is that on the fork as well as Merlin, QoS doesn't have any effect at all. i.e. Speeds don't drop at all (neither ethernet nor wifi) during the dslreports test. This should be a pretty easy scenario to reproduce if you see high bufferbloat on the test. If you do, I don't think the default QoS knob is doing much, if anything at all, to counter it.