Asuswrt-Merlin 376.44 Beta 1 is out

[RMerlin]

With beta3 on the RT-AC87R I noticed the jffs was getting a copy of syslog from the /tmp directory. Any idea of what process is doing this? The syslogd startup shows this:

Asus has been storing Syslog in /jffs for a few months now.

On the entware side opkg returns an
error.

Entware is not supported on ARM-based routers, as the repo only contains packages compiled for the MIPS CPU architecture.

 

[kohht]

You need to check your filesystem permissions through SSH.

Can you provide me the guide regard on filesystem permission? Thanks.

 

[Sushiglobster]

Use native. It will pick up prefix delegation and propagate it to your lan. Also QOS was a no go in with IPv6 turned up on Tomato RAF. I have since shutdown QOS here also.

Good news! Adjusting MTU to 1472 did the trick and I'm no longer getting just 5Mbps. I'm hitting 20Mbps now.

My question, is there anyway to improve this speed? Can I manually adjust or fine tune the MTU?

I get about 45Mbps download with the IPv4 speed test, but with IPv6, I'm getting 20...so about half of my maximum IPv4 speed.

Any suggestions?

Edit: screenshot -

 

[sinshiva]

Good news! Adjusting MTU to 1472 did the trick and I'm no longer getting just 5Mbps. I'm hitting 20Mbps now.

My question, is there anyway to improve this speed? Can I manually adjust or fine tune the MTU?

I get about 45Mbps download with the IPv4 speed test, but with IPv6, I'm getting 20...so about half of my maximum IPv4 speed.

Any suggestions?

Edit: screenshot -

i use TCPMSS and get nearly 10/10mbit down. 60 less than your ipv6 mtu. in your case (replace INTERFACE with what you're using (ifconfig));

ip6tables -I FORWARD -o INTERFACE -p tcp --tcp-flags SYN,RST SYN -j TCPMSS --set-mss 1412

an aside, if you aren't using PPPoE, then an MTU of 1480 with --set-mss 1420 should work

 

[Sushiglobster]

i use TCPMSS and get nearly 10/10mbit down. 60 less than your ipv6 mtu. in your case (replace INTERFACE with what you're using (ifconfig));

ip6tables -I FORWARD -o INTERFACE -p tcp --tcp-flags SYN,RST SYN -j TCPMSS --set-mss 1412

an aside, if you aren't using PPPoE, then an MTU of 1480 with --set-mss 1420 should work

Thanks for the reply. I'm not quite well versed in such stuff. I honestly have no idea what you're talking about, haha. I'm not sure where to input those commands.

My speed slightly improved. See the pic at the bottom. I actually followed a guide specifically for my ATT UVERSE service that deals with cascading a router (http://forums.att.com/t5/Features-a...n-bridge-mode-or-as-close-as-you/td-p/3552057). Disabling my ATT gateway's firewall and setting it into passthrough mode so the RT-N66U could handle everything helped increased IPv6 speeds.

However, I'm still at about half of IPv4 speeds. Not sure what else I can do...???

Pic:

Anyways, not sure if it's a IPv6 issue with the firmware being BETA and all or if it's because I'm tunneling with IPv6 (Tunnel 6rd) and thus it will always be slower because of this. I'm not an expert or anything so I really have no idea.

 

[Techno Pride]

Ok, managed to fix the AC66U cfe problem. It now has SG settings.

Tried the beta3 fw and did a full reset via the UI.

Noted the following:

Doesn't work
Wireless Mode: N+AC
Channel bandwidth: 20/40/80 MHz
Control Channel: 100

Works:
Wireless Mode: Auto
Channel bandwidth: 20/40/80 MHz
Control Channel: 100

Tx power seems to be limied to 100:

admin@ac66u:/tmp/home/root# nvram show | grep power
size: 42193 bytes (23343 left)
Ate_power_on_off_ver=2.4
wl_txpower=100
wl0_txpower=100
pci/2/1/measpower1=0x7f
pci/2/1/measpower2=0x7f
Ate_power_on_off_enable=0
pci/2/1/measpower=0x7f
wl1_txpower=100

 

[majnu]

Works for me. My Nexus 7 shows as a tablet. Could be because your client list hasn't fully refreshed/rescanned.



Must be some kind of hybrid there. Is it an AC66 or a N66?

Sorry. AC66u. But just checked my router after letting it run with the fw overnight and it's sorted itself out.

And still you haven't answered RMerlin's question about which router you have.

 

[majnu]

Ok, managed to fix the AC66U cfe problem. It now has SG settings.

Tried the beta3 fw and did a full reset via the UI.

Noted the following:



Tx power seems to be limied to 100:

Beta 2 and 3 both have been like this for tx power.

 

[Ronv42]

Connect to DNS server in IPv6 settings

The toggle is backwards on this option for the AC87:

Connect to DNS Server automatically

If you say yes there are no fields for a IPv6 DNS server. If you say no the DNS server fields show up.

 

[jim769]

@ Merlin any idea when there will be a 376.44 final ? I would like to give it a try.

 

[terribell]

Wifi performance on the RT-N66U

I've been using Merlin for a long time and it has always been a big stepup from the official fw, but this latest .44 release has degraded greatly the wireless performance, even though they corrected some problems I've been having with DLNA.

Today I've been trying several firmwares, with reset of the configuration and removal of wifi profiles and the last ones that seem okay are .42 and .43, I've also tested with INSSIDer and also throughtput doing copies of files.

With all the previous firmwares I could get HD on the chromecast I have in the bedroom, but with .44 beta 1, beta 2 and beta 3 I can only get SD and always with cuts, so it's impossible to watch.
With a win7 notebook on the chromecast location I can check a 10db loss after the change in fw from .43 to any of the .44 betas.

I know that Merlin is sick of people complaining about wifi performance and decided to go with the official and current ASUS driver but I would like to appeal to one last chance to try and keep the old driver.
If this is not the case unfortunatelly I will not be able to upgrade as it breaks some devices connection where they are more distant from the router, and will have to keep broken DLNA service.

Thanks for all the hard work and understanding my appeal.

 

[Ronv42]

i use TCPMSS and get nearly 10/10mbit down. 60 less than your ipv6 mtu. in your case (replace INTERFACE with what you're using (ifconfig));

ip6tables -I FORWARD -o INTERFACE -p tcp --tcp-flags SYN,RST SYN -j TCPMSS --set-mss 1412

an aside, if you aren't using PPPoE, then an MTU of 1480 with --set-mss 1420 should work

Thanks for this one. Fixed issues I was having with Xbox Live and Facebook. Here is the command I used:

ip6tables -I FORWARD -o vlan2 -p tcp --tcp-flags SYN,RST SYN -j TCPMSS --set-mss 1412

 

[bodean]

Any way to edit the TOPIC TITLE to reflect we are on Beta 3?

 

[RMerlin]

Can you provide me the guide regard on filesystem permission? Thanks.

Lots of info on the web on how to use chmod. The short version:

chmod 777 -R /mnt/sda1

(assuming disk is mounted as sda1)

 

[RMerlin]

The toggle is backwards on this option for the AC87:



If you say yes there are no fields for a IPv6 DNS server. If you say no the DNS server fields show up.

If you say No, then it won't be automatic, meaning you get fields to enter manual options. That looks correct to me.

 

[RMerlin]

Any way to edit the TOPIC TITLE to reflect we are on Beta 3?

I stopped doing that a few months ago since it breaks things for people subscribed to the title to be notified of new posts.

 

[PabloAbonia]

The kernel options are shared amongst all ARM devices, which is why IPSEC modules are disabled for all three of them. You will have to modify kernel options and recompile the firmware if you want to enable IPsec support.



Plus, once Asus implements DPI support on the other ARM devices (I suspect they eventually will), you will be back to square one, so best get used to it already.

You may be misunderstanding me. I did not recompile it with with any BWDPI changes to target.mak, and the option was set in a manner that would allow for the use of IPSEC. I used the firmware you compiled, and I was assuming that what I downloaded to do my own compile was reflective of your compile firmware. I'm still able to use IPSEC with the firmware you compiled.

As far as the DHCPv6 flipping of bits is concerned, I realized that having both router advertisement and enabling the DHCPv6 server resulting in starting both radvd and dhcpv6s, causing a battle for supremacy between these ipv6 components. Unless all of their configuration is aligned exactly, they will try to override each other with dueling router advertisements. Using ipv6 tools, it could occasionally capture two different ipv6 addresses to be used for the router.

Assuming this is correct, I think it would make sense to make them mutually exclusive under most circumstances.

Pablo

 

[bluepoint]

I've been using Merlin for a long time and it has always been a big stepup from the official fw, but this latest .44 release has degraded greatly the wireless performance, Thanks for all the hard work and understanding my appeal.

Yes, same experience. This is the firs time ever I've seen 5GHz hit -60 - 62dB where as it used to be -45 to -50dB where I usually use my laptop. Signal wise according to windows netsh command, is, only @70% as compared to the EM driver up to .43, the wireless signal is 90%, 20 percentage pts. is a huge difference. This 376 wireless driver is even worst than the older non EM SDK6.

 

[Ronv42]

If you say No, then it won't be automatic, meaning you get fields to enter manual options. That looks correct to me.

Seems a little backward maybe the description should be better.

Yes) use current DNS for IPv6.
No) define your own DNS for IPv6.

 

[PabloAbonia]

You may be misunderstanding me. I did not recompile it with with any BWDPI changes to target.mak, and the option was set in a manner that would allow for the use of IPSEC. I used the firmware you compiled, and I was assuming that what I downloaded to do my own compile was reflective of your compile firmware. I'm still able to use IPSEC with the firmware you compiled.

As far as the DHCPv6 flipping of bits is concerned, I realized that having both router advertisement and enabling the DHCPv6 server resulting in starting both radvd and dhcpv6s, causing a battle for supremacy between these ipv6 components. Unless all of their configuration is aligned exactly, they will try to override each other with dueling router advertisements. Using ipv6 tools, it could occasionally capture two different ipv6 addresses to be used for the router.

Assuming this is correct, I think it would make sense to make them mutually exclusive under most circumstances.

Pablo

After performing a packet capture, I've discovered that a group of my home automation devices are responsible for the odd ball ipv6 router announcements. I deal with that with the company involved.

Regardless, I don't think having both running simultaneously on the router (radvd and dhcp6s) would be a good idea.

Thanks

Pablo