[Release 384/NG] Asuswrt-Merlin 384.5 is now available

[rob-tech]

Only the newest releases mention fixed ipv6 issues. Merlin, As native ipv6 seems broken with 384.5, I think you should wait for 3.0.0.4.382.50624 and 3.0.0.4.384.21045.

 

[Morac]

I have no idea what happened, but today when I got home my Internet was spotty. At first I thought it was my ISP, but I checked my router logs and it wad reporting over and over that it couldn’t write to /var/lib/misc/dnsmasq.leases because there was no space left on the device. This went back as far as the log went to June 11 8:58 pm, so nearly 24 hours.

I have a AC88U which has about 500 MB of ram of which normally a little over 320 MB is free. Before rebooting I had 75 MB free. I wanted to ssh in to the router to see what was using all the space, but there wasn’t enough memory to fork a process to do so. All I could do was reboot. I now have 325 MB again.

I don’t run anything special on the router. Just the basics and an OpenVPN server. It had been up 20 days without issue. Any idea what happened?

So I had taken a screenshot of the Tools screen and looked at it a bit closer and noticed that 222 MB of memory was being used for cache. Normally that’s around 12 MB, so the router was basically giving the cache all the RAM. Isn’t there some kind of limit to prevent that?

 

[scjr]

So I had taken a screenshot of the Tools screen and looked at it a bit closer and noticed that 222 MB of memory was being used for cache. Normally that’s around 12 MB, so the router was basically giving the cache all the RAM. Isn’t there some kind of limit to prevent that?

Under Tools/Other Settings under Advanced Tweaks, is Memory Management set to Yes or No? If set to No, it will use more memory. It allows the system to manage the memory. At least, that’s what I seen posted here by @skeal and others. Not sure if that’s what it is, but worth a look.

 

[Morac]

Under Tools/Other Settings under Advanced Tweaks, is Memory Management set to Yes or No? If set to No, it will use more memory. It allows the system to manage the memory. At least, that’s what I seen posted here by @skeal and others. Not sure if that’s what it is, but worth a look.

It is set to yes, but I’m guessing that’s done via a cron job or something. Cron jobs were failing because the router didn’t have enough memory to fork processes. I still don’t know why the cache size was 200% of what it normally is. I don’t do anything fancy on it like run programs. I just use it as a router.

 

[Treadler]

Only the newest releases mention fixed ipv6 issues. Merlin, As native ipv6 seems broken with 384.5, I think you should wait for 3.0.0.4.382.50624 and 3.0.0.4.384.21045.

Working ok here.
I’ve experienced occasional glitches on 384.5 though.
I tried the OE firmware, 21045, for a couple of days, yes indeed, the IPv6 appears totally stable on it.

 

[rob-tech]

Working ok here.
I’ve experienced occasional glitches on 384.5 though.
I tried the OE firmware, 21045, for a couple of days, yes indeed, the IPv6 appears totally stable on it.

Yes, this is what I mean. I used the latest stock firmware before and ipv6 was fully stable. Right now with 384.5 the clients may have working ipv6 for only a few minutes after connecting before they drop the WAN address, while only retaining the link-local. I have a simple PPPoE VDSL with the modem bridged so the Asus does all the work.

This is obviously no fault of Merlin if Asus' code regressed. However, the simple reason that they mention ipv6 fixes and my observations (along with yours) means that it's worth looking into, as this is core functionality. It would be a shame if Merlin does the hard work of merging the latest code and a major feature remains regressed for another cycle.

 

[DJCrapsody]

GPL for 3.0.0.4.384.20942 is out. However, IPv6 still not fixed for this release, so we gotta wait a little longer for GPL for 3.0.0.4.384.21045 to be released.

 

[kfp]

There are 384.6 alpha builds if you’re willing to try, seems to include some IPv6 fixes but not 100% if these are the same patches in 21045.

Mostly bug fixes so should be relatively stable.

 

[Deplorable]

Samba performance is a third of what it was before upgrading. Definitely not a keeper.

 

[Treadler]

There are 384.6 alpha builds if you’re willing to try, seems to include some IPv6 fixes but not 100% if these are the same patches in 21045.

Mostly bug fixes so should be relatively stable.

Ha ha, good catch! I hadn’t noticed the alpha.
I might have to give it a try!

 

[rob-tech]

There are 384.6 alpha builds if you’re willing to try, seems to include some IPv6 fixes but not 100% if these are the same patches in 21045.

Mostly bug fixes so should be relatively stable.

I'll wait for the next stable version, ipv6 is nice but it's not worth flashing an alpha over this 384.5 version which is essentially flawless, apart from ipv6 this build provided over 17 days of problem free service so far (Adaptive QOS and all protection features enabled, the QOS in particular is a great improvement over stock with my 15/10 Mbps VDSL).

 

[unsynaps]

GPL for 3.0.0.4.384.20942 is out. However, IPv6 still not fixed for this release, so we gotta wait a litle longer for GPL for 3.0.0.4.384.21045 to be released.

What is broken in IPv6? Works fine here.

 

[Morac]

What is broken in IPv6? Works fine here.

Same for me.

 

[scjr]

Same for me.

Works fine for me as well on 384.5.

Changelog for 384.6 alpha_1 mentions some fixes for ipv6.

 

[Makaveli]

What is broken in IPv6? Works fine here.

Robtech's post mentions the issue.

I had it happen to me when I first loaded the firmware.

Any test from http://ipv6-test.com/

would fail.

A reboot of the router and redoing the IPv6 setup solved the issue for me it has not come back since.

 

[Bamsefar]

I think I might have solved my MAC address filter issue on my router, since it now works 100%...

Now it would be nice if I could say why, but I can't. What I have done is about 10 reboots, removed my Nexus 7 (2013) from 5GHz (moved it to 2.4GHz only), and most importent did not add any new MAC addresses in 5 minutes after last added, and to top it off added WiFi manually on my 3 laptops all running Windows 10 (latest). So not sure why it is stable now, but it is and I will try to not touch my router...

...and knowing me that will last for about a day or so...

 

[strumf666]

I just checked system log and it's full of messages like this:

Jun 18 02:46:21 kernel: br0: received packet on vlan4000 with own address as source address
Jun 18 02:46:22 kernel: br0: received packet on vlan4000 with own address as source address
Jun 18 02:46:24 kernel: br0: received packet on vlan4000 with own address as source address
Jun 18 02:46:25 kernel: br0: received packet on vlan4000 with own address as source address
Jun 18 02:46:26 kernel: br0: received packet on vlan4000 with own address as source address
Jun 18 02:52:10 kernel: br0: received packet on vlan4000 with own address as source address
Jun 18 03:11:54 kernel: br0: received packet on vlan4000 with own address as source address
Jun 18 03:11:55 kernel: br0: received packet on vlan4000 with own address as source address

What does this mean?

 

[ech]

I have no idea what happened, but today when I got home my Internet was spotty. At first I thought it was my ISP, but I checked my router logs and it wad reporting over and over that it couldn’t write to /var/lib/misc/dnsmasq.leases because there was no space left on the device. This went back as far as the log went to June 11 8:58 pm, so nearly 24 hours.

I have a AC88U which has about 500 MB of ram of which normally a little over 320 MB is free. Before rebooting I had 75 MB free. I wanted to ssh in to the router to see what was using all the space, but there wasn’t enough memory to fork a process to do so. All I could do was reboot. I now have 325 MB again.

I don’t run anything special on the router. Just the basics and an OpenVPN server. It had been up 20 days without issue. Any idea what happened?

I just had OOM kill the web server on my router, after running for 21 days w/o problems... seems something is leaking memory. And I notice that just enabling the AiProtection chews up 60Mb or so of my 256Mb of memory right away.

 

[guho]

On my AC3200 with 384.5 I had lots of trouble connecting wifi devices on the 2.4GHz band (have separate SSIDs for 2.4, 5-1 & 5-2). Downgrading to 384.3 did not help either. Fiddled with Professional 2.4GHz settings, turned off airtime fairness etc. to no avail.

In despair, I flashed the latest Asus stock firmware on top of Merlin 384.5 and did not do a factory reset (because many posters on this forum indicated their 2.4GHz problems did not get resolved by resetting NVRAM). All my settings remain (of course lost Merlin features) and all my 2.4GHz WLAN devices connect fine! I hope there will be a Merlin version soon that has the opensourced Asus code corresponding to the code in the current stock Asus Ac3200 firmware (released 6/1/18). Then I will happily switch back to Merlin because I need OpenVPN Server and the JFFS scripting to tweak my guest network.

 

[agilani]

In case it helps anyone else - after struggling with wireless issues with the new code base and constant reboots and even trying stock. I went through every option including disabling mimo and beam forming and even trying to get a hold of asus support. This was my personal favorite from asus support, "Please be reminded that we can only communicate with our software engineer via email on our end. We can assure you that once we have forward the information that you will provide for your router our engineers can further replicate and further rectify the concern."

Just as i was ready to go back the old code base, i noticed that every time i had the problem wireless debug showed the router using control channel 5. I changed it to 7 and so far the router has been up for over 3+ days without any 2.4 ghz issues. Still watching it to see if anything changes, but its been stable so far. I have also noticed the SNR and RSSI has always been 0 which is strange. It can't be that simple can it? maybe the new code base doesn't calculate interference when selecting channel? why else would rssi and snr both be 0?