What's new

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

  • SNBForums Code of Conduct

    SNBForums is a community for everyone, no matter what their level of experience.

    Please be tolerant and patient of others, especially newcomers. We are all here to share and learn!

    The rules are simple: Be patient, be nice, be helpful or be gone!

scp encrypts all traffic, so devices with slow processors will struggle, which may be part of the problem. Maybe try some unencrypted protocols like FTP or NFS?

These are non-public personal data so unfortunately I can't do unencrypted transfers.

I think my big surprise is from the difference between the WAN and LAN bandwidth when using the same encrypted mechanism (SCP) on the same device: 270KB/s vs 1.5MB/s. That is what makes me wonder if there is a config or firmware issue.
 
Last edited:
+1 on this....I just ran a test across the local lan copying from PC to USB3 key, AC68P, 4GB file, NTFS....
Using SCP (via WINSCP) write ~6MB/s
Using SMB (via Windows Copy) write ~40MB/s

If you do a google search, you'll find multiple references to slow SCP performance.

Do you think if SFTP may do better than SCP? Just wondering if you guys know, before I spend bunch of time manually setting it up to test it (planning to start with the configs at http://www.snbforums.com/threads/vsftpd-support-ssl-tls-encryption-support.17939/#post-154537 and assuming that related changes by Merlin were ported to this fork :))
 
Do you think if SFTP may do better than SCP? Just wondering if you guys know, before I spend bunch of time manually setting it up to test it (planning to start with the configs at http://www.snbforums.com/threads/vsftpd-support-ssl-tls-encryption-support.17939/#post-154537 and assuming that related changes by Merlin were ported to this fork :))

Both SFTP and SCP use SSH, so you should get the same performance.

FTPS (FTP with TLS/SSL) is another topic, but it's likely to perform just as bad.


These routers are just not powerful enough to deal with much encryption. For better performance you could choose a different, less CPU-intensive encryption cipher or run OpenSSH or VSFTPd on a desktop PC which has much more powerful CPUs.


You can always try a router reset. Check top/htop first to see what your CPU usage is during a file-transfer.
 
You could try something like this:

rsync -rtvv --stats --progress -e "ssh -A -o SeverAliveInterval=15 -t admin@ROUTERIP ssh -A username@192.168.1.2" /media/folder :/Data/username

It works like this with a RT-N16 and a Seagate Central NAS, speed is between 500 kB/s and 1 MB/s.

By the way, scp-ing to the hard disk behind the router, I get 800 kB/s to 1 MB/s. So your connection may play a role.

Router does need then a swap partition (mounted, too).
 
what does speed up SCP and SFTP is if sshd has the hpn feature patched in. Whether or not John can do this on a embedded OS tho I have no idea. Otherwise ssh is capped to a small TCP buffer size which restricts speeds.

https://www.psc.edu/index.php/hpn-ssh

That project seems to focus on high-bandwidth devices with ample processing power where undersized buffers become the primary throughput bottleneck. We are limited by the CPU.

For example, the slowest speed in their graph is faster than the fastest speed of the fastest Asus router is capable of.
 
Do you think if SFTP may do better than SCP? Just wondering if you guys know, before I spend bunch of time manually setting it up to test it (planning to start with the configs at http://www.snbforums.com/threads/vsftpd-support-ssl-tls-encryption-support.17939/#post-154537 and assuming that related changes by Merlin were ported to this fork :))
I suspect that this may do better as it uses OpenSSL for the encryption, but I really don't know by how much. I wouldn't expect miracles, however....maybe get the your WAN number up closer to 1MB/s.

To answer your second question, I hadn't touched vsftpd at all since the beginning of the fork. I just added the Merlin changes to to support sftp/openssl for the next release.
 
John, I've been seeing this random reboot on my RT-AC66R, any idea what is causing this? I thought it could be the USB but while doing the slow version of restoring to factory defaults, it still continued without the drive connected. Could the master router be causing this? The log should be attached

couldn't take the full log, but attached
 

Attachments

  • short.txt
    61.7 KB · Views: 582
Last edited:
That project seems to focus on high-bandwidth devices with ample processing power where undersized buffers become the primary throughput bottleneck. We are limited by the CPU.

For example, the slowest speed in their graph is faster than the fastest speed of the fastest Asus router is capable of.

not really, tcp buffers can have a huge affect, it all depends on the RTT.

e.g. from eu to usa, with a small rwin, you wont get much more than a few mbit/sec at best, the kind of situation the HPN patch is aimed at. Thats easily doable on a router.

Sadly john has confirmed he cannot add hpn due to not using openssh.
 
not really, tcp buffers can have a huge affect, it all depends on the RTT.

e.g. from eu to usa, with a small rwin, you wont get much more than a few mbit/sec at best, the kind of situation the HPN patch is aimed at. Thats easily doable on a router.

Sadly john has confirmed he cannot add hpn due to not using openssh.

I suspect my much lower WAN speed might have been related to the connection latency causing issues with the way the buffer works and what HPN is trying to address. Unlike my lan traffic, the WAN speed was for data being written from the west coast to the router on the east coast.

My workaround setup for the ssh limits is now to have an openssh server on a pc on the lan (w/ port forwarding at the router) to transfer the data over sftp. and then use smb for getting the data on the hdd connected to the router. With the same west coast-east coast connection, I am maxing the ISP's bandwidth limit. I am not sure if the latest version of openssh for windows which I am using has the HPN patch or not thou, so maybe the latency isn't as much of an issue. Having a two step process instead of a one step setup-and-forget-it process is somewhat of a pain thou.
 
Hello john,

first of all I want to thank you for the images you offer here. I'm using them on both of my router (AC68P and N66U) and they run very stable.
There's just on wish I have. I'm planning to use the AC68P as TimeMachine-Server for my iMac and my MacbookAir. While checking the status of netatalk with afpd -V the result is that version 3.0.5 is installed. According to the website of netatalk there is v 3.1.0 available. Would it possible to upgrade to the latest version with one of your next releases?
That would be great.

Thanks in advance
 
@Santiago C - By this I meant I tried WAN+USB and WAN+Ethernet LAN (I used LAN Port 4). Both worked OK.

Next thing for you to try is to enable the browser debugger (F12) and check for any big errors under 'Console' (there are always some 'noise' errors).

Hi @john9527. Nothing weird on the debugger, no messages whatsoever and I just had the chance to test it again, starting from scratch (installed 20E3, did both WPS button and GUI resets), put my WAN in PPPOE mode, activated Dual WAN with LAN1 and upon restart while I do see the dropdow with "WAN" and "Ethernet LAN" nothing happens when I switch back and forth ("Ethernet LAN" should be configured with Automatic IP if I'm not mistaken, but fields do not get refreshed)
Now, let's say the display is stuck in "WAN", if I got to "USB Application" - "3G/4G" that drives me to Advanced_Modem_Content.asp where, even if it displays the 3G modem config settings, if I switch from "WAN" to "Ethernet LAN" then I am able to see the LAN1 config but not go back (everytime I get to Advanced_WAN_Content.asp I would see the LAN1 config, unless I do this whole "procedure" again to get the WAN config).
I though I might have had some weird configuration preventing me from doing this, but having tested it in a clean configuration really puzzles me...

On a second review, here's what I found:
Line 105 of Advanced_WAN_Content.asp (function change_wan_unit(obj)) does a "return false;" if switching to a "WAN" or "Ethernet LAN" entry and that prevents execution of the following piece of code:
HTML:
FormActions("apply.cgi", "change_wan_unit", "", "");
document.form.target = "";
document.form.submit();
Code that, if put in the browser console (after having a breakpoint in that "return false;" line) does exactly what I expect.

In the current Merlin version that "return false;" seems to be replaced by:
HTML:
if(obj.selectedIndex != wan_unit_flag){
                                document.form.wan_unit.value = obj.selectedIndex;
                        }      
                        else{
                                return false;
                        }
I guess that code only returns false if the selected entry is the initial one, am I right?

Regards!
 
I guess that code only returns false if the selected entry is the initial one, am I right?
Sorry...next contestant please :) Just kidding.....
The change in Merlin is common code that was picked up in his merge that supports the next gen router which has 2 WAN ports.

But, good explanation and debug none-the-less. I now see what you mean and duplicated it. They were making an assumption for the gui that both dual wan connections would have the same configuration for WAN/Ethernet dualwan. I checked, and there are independent settings for each connection, so hopefully it's not a restriction that's enforced somewhere else (and the same gui assumption is still in the latest Merlin). I'll make the change to make them independent, and hopefully it will work.
 
Sorry...next contestant please :) Just kidding.....
The change in Merlin is common code that was picked up in his merge that supports the next gen router which has 2 WAN ports.

But, good explanation and debug none-the-less. I now see what you mean and duplicated it. They were making an assumption for the gui that both dual wan connections would have the same configuration for WAN/Ethernet dualwan. I checked, and there are independent settings for each connection, so hopefully it's not a restriction that's enforced somewhere else (and the same gui assumption is still in the latest Merlin). I'll make the change to make them independent, and hopefully it will work.

Well, I am not well versed in that kind of code and debugging, so I did my best ;)
Even with that assumption, if you change the WAN unit and then proceed to make further changes, those changes would be applied to the first selected entry since the form has not been resubmitted, am I right? (gotta get one answer right sometime)

Another thing I found is that if you choose not to connect automatically to DNS (in order to display the manual fields and complete them), changes are saved but whenever I go back to that page the DNS setting is in “Yes” although it is still using the manual entries (I use OpenDNS and can see in the dashboard that requests are being placed)

Regards!
 
I haven't posted about this without testing several firmware revisions but at least on my particular RT-N66U any build newer than 17E8 creates problems using the 2.4ghz radio (constant buffering or connectivity intermittently unavailable), switching to 5ghz radio works perfectly. Any ideas? Using 20mhz on 2.4ghz, ch1, WPA2/AES never once had any problems on 17E8 or older, each time I try newer I immediately notice it with slacker radio for instance.

I have also reset defaults etc troubleshooting the newer builds, makes no difference, always end up back on 17e8
 
particular RT-N66U any build newer than 17E8 creates problems using the 2.4ghz radio
There was a change in that time frame, but it was an nvram only option to enable dfs channels on the 5GHz band for EU N66's. Nothing was changed on the 2.4GHz radios.

I double checked the code and had someone I work with on testing dump the 2.4GHz parameters and confirmed no change on the 2.4GHz radio.

To double check on your router, please run the following command via telnet/ssh on both the firmware you believe is working and when you believe it's not. The results should be identical.
Code:
nvram show | sort | grep -E 'regulation_domain|country|regrev|reg_mode|ccode|dfs_enable'
 
Hi @john9527 and all! Is there any easy way to assign static IP addresses to all clients (I have about 50 of them)? For example, supplying a text file with all the MAC, desired IPs & names to a script that can be executed? Thanks!
 
Update, please, DLNA assembly. With the ability to save space stop video playback. Thank you.
P.S. google translate.:oops:

That's the job of the client, not the DLNA server.
 

Sign Up For SNBForums Daily Digest

Get an update of what's new every day delivered to your mailbox. Sign up here!
Top