Asuswrt-Merlin 378.56_2 is now available

[pushkinsk]

Just updated my RT-N66U. The MAC filter has been split off into 2.4ghz, and 5ghz like it was previously. This causes some issues when copying the configuration over from the previous 378.55 build. It adds the name of the Device on the end of the MAC like this AA:AA:AA:AA:AA>Name. It won't allow you to delete the entries, and adds garbage to the end after each delete, and apply attempt. To restore the configuration I did a factory reset, then used the NVRAM save and restore tool from these forums. I also tried restoring from the UI, which gave the same results.

I had the same issue.

Solution:
1) backup macfilter in ASCII format (TXT)
2) activate ssh daemon
3) login
4) nvram show| grep maclist
5) clear all macfilters for all wlan interaces ("nvram set wl[|0|1]_maclist[|_x]=")
6) nvram commit
7) reenter all required macfilters from GUI

 

[pushkinsk]

My RT-AC87U router has Quantenna firmware version 37.3.0.50. Which version is the latest?

 

[HallStevenson]

Also, the usual: reboot with no USB disk plugged in, then try again to upgrade.

I may have read it in the previous thread on the 1st beta that resetting to factory defaults should be done before flashing. Is that recommended, required, or what would people suggest. I have Merlin's previous "official" release installed currently.

I know that if I "dirty" flash and have issues that I better "clean" flash before reporting anything here !

 

[Blinkyz]

My RT-AC87U router has Quantenna firmware version 37.3.0.50. Which version is the latest?

Stock Asus = 3.0.0.4.378.9177
https://www.asus.com/ca-en/Networking/RTAC87U/HelpDesk_Download/
Merlin = 378.56_2 ( link at the beginning of this thread )

 

[ajoyner]

Thanks pushkinsk. I didn't use your code verbatim, but it got me on the right track. I used:

1. nvram show | grep maclist
2. copy the text of wl0, and wl1 and paste to textpad
3. remove garbage characters from the mac addresses.
4. nvram set wl_maclist_x="corrected mac's here"
5. nvram set wl0_maclist_x="corrected mac's here"
6. nvram set wl1_maclist_x="corrected mac's here"
7. nvram commit
8. service restart_wireless

 

[greggy101]

My RT-AC87U router has Quantenna firmware version 37.3.0.50. Which version is the latest?

I have the same.

 

[Joel Teixeira]

Hi guys, I believe my question is not specific for this release and I couldn't find it in other topics so sorry if it's already been asked. What file should I edit to set aliases?

I used entware.arm/etc/profile but it doesn't seem to work.

Also, trying to change the default shell from sh to bash I followed this guide:

http://www.snbforums.com/threads/bash-history.16694/#post-115010

But a echo $Shell still showing /bin/sh. Is is info outdated? Thanks.

Sent from my SM-N910C using Tapatalk

 

[mromero]

On the RT AC-87U Is there anything really useful or practical in this reversion apart from calming down the Nervous Nellies?

Asuswrt-Merlin is now available for all supported models, including the RT-N66U build.

378.56_2 (2-Nov-2015)
    - CHANGED: Reverted the memory buffering optimization
               for ARM devices, as people keep panicking
               over the lower amount of free RAM.  You can
               manually re-enable the optimization by setting
               "drop_caches=0" in nvram.
[/QUOTE]

 

[RMerlin]

I have the same.

The Quantenna firmware is part of the main firmware. The version will always be the same for everyone running the same version of Asuswrt...

 

[RMerlin]

On the RT AC-87U Is there anything really useful or practical in this reversion apart from calming down the Nervous Nellies?

Not that I know.

 

[mromero]

Not that I know.

And are there any useful or practical performance benefits to going back to your memory optimization capabilities using the instructions posted?

 

[RMerlin]

And are there any useful or practical performance benefits to going back to your memory optimization capabilities using the instructions posted?

Some people mentioned that the web interface was faster for them. Personally I couldn't really see any difference.

 

[john9527]

And are there any useful or practical performance benefits to going back to your memory optimization capabilities using the instructions posted?

When we did the development on my fork, the user who noticed the difference between his previous MIPS based router (don't remember if it was n N66 or AC66) and AC68 had scenarios where the AC68 would perform worse. The MIPS routers don't do the dropcache. Implementing the no-drop-caches fixed the AC68. Maybe @Chrysalis can comment further.

 

[john9527]

It's been a while, but I seem to remember he had a 'flakey' ISP, that would cause a lot of error recovery. It severely impacted the AC68 without the no-drop-caches.

 

[Luis Ibarra Consiglieri]

Hi Merlin, Hi guys...

I have a AC3200 with the 378.56_2 in Access Point mode and the Traffic Analyzer/Statistic doesn't work, and in the System log appear these problems:

Nov 5 02:00:17 hour monitor: dpi engine doesn't exist
Nov 5 02:00:27 hour monitor: dpi engine doesn't exist
Nov 5 02:00:37 hour monitor: dpi engine doesn't exist
Nov 5 02:00:47 hour monitor: dpi engine doesn't exist
Nov 5 02:00:57 hour monitor: dpi engine doesn't exist
Nov 5 02:01:07 hour monitor: dpi engine doesn't exist
Nov 5 02:01:17 hour monitor: dpi engine doesn't exist
Nov 5 02:02:17 hour monitor: dpi engine doesn't exist
Nov 5 02:03:17 hour monitor: dpi engine doesn't exist

Is this an error? is posible get the Traffic Analyzer/Statistic option in AP mode?


PD: the 378.56_2 version is most stable than 378.56 I think because:
- CHANGED: Reverted the memory buffering optimization
for ARM devices, as people keep panickingn
over the lower amount of free RAM. You can
manually re-enable the optimization by setting
"drop_caches=0" in nvram.
- FIXED: Debug log from mDNSNetMonitor could gradually
fill up RAM - disabled it.

Thanks in Advance.

 

[Mattrazol]

Asuswrt-Merlin is now available for all supported models, including the RT-N66U build.

Summary of changes since 378.55:

• Merged with Asus GPL 378_9177, which includes numerous fixes and tweaks, and improvements to the networkmap. This includes the newer Quantenna driver for the RT-AC87U, which resolves connectivity issues on the 5 GHz band, however it still causes some battery drain on that band. The only workaround is to revert to 378.55 if this is a problem for you.
• Support was added for the (unreleased yet) RT-AC88U. Thanks to Asus for providing a development sample.
• Enhancements to syslog handling. The default log level can now be configured (it used to be hardcoded to Emergency). This is especially useful to users who log to a remote syslog server.
• Editable entries for the DHCP static lease, virtual servers and prot triggers pages.
• Various component updates: miniupnpd, openvpn, dropbear and minidlna were updated to their latest versions.
• Various issues related to networkmap were resolved, especially handling of long hostnames.
• You can now configure up to five separate OpenVPN clients, to match Asus's stock FW.
• Numerous other fixes, please see the complete changelog for details

Additional changes in 378.56_2:

378.56_2 (2-Nov-2015)
    - CHANGED: Reverted the memory buffering optimization
               for ARM devices, as people keep panicking
               over the lower amount of free RAM.  You can
               manually re-enable the optimization by setting
               "drop_caches=0" in nvram.
    - CHANGED: Allow using a port < 1024 for http(s) webui
               interface.
    - FIXED: EMF wasn't working on AC56/AC68/AC87.
    - FIXED: Couldn't connect to ISPs using VLANs (RT-AC87U)
    - FIXED: Editing Port Forward entry with ellipsis in
             the description or the port range would
             still edit the shortened version instead
             of the full content.
    - FIXED: Debug log from mDNSNetMonitor could gradually
             fill up RAM - disabled it.
    - FIXED: Router crash if pasting SSH key > 2047
             characters.
    - FIXED: Editing an entry on the networkmap would
             clear the hostname if entry existed in
             the DHCP static list.
    - FIXED: OpenVPN server in secret key mode
             would fail to start.
    - FIXED: Couldn't add entries to the MAC Filter list
             of Guest Networks (reverted our previous
             implementation which conflicted with
             Asus's new one).
    - FIXED: NTP failing to refresh for some cases.
             Implemented temporary workaround.
    - FIXED: Some services not properly starting at
             boot time (like Parental Control or Tor)

Downloads are here.
Changelog is here.

Hello i'm Matt I'm new here! I would like to know from you RMerlin which setting is the best for optimum performance? drop_caches=0 or 1? What are the main differences? How are the cache drops managed? And if I'll update with future firmwares do I have to set the value after every update? Thanks for the reply!

 

[charlie2alpha]

Hello i'm Matt I'm new here! I would like to know from you RMerlin which setting is the best for optimum performance? drop_caches=0 or 1? What are the main differences? How are the cache drops managed? And if I'll update with future firmwares do I have to set the value after every update? Thanks for the reply!

When that setting is set to "1", when services are started or restarted it sends "1" to /proc/sys/vm/drop_caches. What it does you can read here.

 

[larsnet]

This appears to be a known issue as it was reported with latest Asus official build for RT-N66U and acknowledged by Asus Support. http://www.snbforums.com/threads/official-rt-n66u-firmware-version-3-0-0-4-378-9235.27622/

Thanks for the response! I'll wait for the fix then.

 

[john9527]

When that setting is set to "1", when services are started or restarted it sends "1" to /proc/sys/vm/drop_caches. What it does you can read here.

It's actually executed more often than service start/stop. It's on a watchdog timer that gets executed every 5 minutes.

 

[charlie2alpha]

It's actually executed more often than service start/stop. It's on a watchdog timer that gets executed every 5 minutes.

Didn't know that, thanks for the additional info.