[RT-AC68U] A dose of Quicksilver

[JGrana]

Ok, I didn't return until late Sunday... With respect the the alpha build, but I will run minidlna again with the large Photos directory and report back.
I am going to first run a scan on both the USB 3.0 1.5 TB drive (that holds all the photos, NTFS formatted) and the USB 2.0 drive that is ext3 and holds the /minidlna directory for logs and the DB.

I will also do a few putty sessions with tail -f of minidlna.log and syslog.log

Ok, the good news - the alpha minidlna ran to completion. There was one file (.jpg) it complained about, but it successfully indexed all the files. There were over 55000 photos.
The bad news - after it was done I couldn't access the files/server. I use both a Samsung TV and an Android app. Neither saw the AC68U as a dlna server. I tried a reboot - unfortunately the router only mounted one of the two attached USB drives. It mounted the USB 3.0 drive but did not mount the smaller one connected to USB 2.0. This happens occasionally. The USB 2.0 drive has the database on it...

So, it looks like the memory leaks are gone. Now, I believe all we need is for Asus to firm up the USB software and we should be in good shape.

Merlin - thanks for the fine work!

BTW - it ran noticeably faster then the AC66U I have - same disks!

 

[RMerlin]

So, it looks like the memory leaks are gone. Now, I believe all we need is for Asus to firm up the USB software and we should be in good shape.

Merlin - thanks for the fine work!

Thanks for confirming it Paulo did a pretty good job at tracking down all those leaks and fixing them one by one.

BTW - it ran noticeably faster then the AC66U I have - same disks!

The big limitation there is the disk IO performance - something a USB3 disk should do much better than the USB2 disk on an AC66. I would expect performance to be around 3-4x faster with the AC68U in this area.

 

[bigcid10]

well this firmware is working ok on my ac68
loopback is working fine as well as port forwarding
it also doesn't crash either like 217 and 235,and 291 did

 

[canuckle]

How is the N performance on both bands? Can I use this firmware as a reliable platform? Anything broken in port fwd or manual ips? My needs are basic, no VPN or attached USB drives, no aicloud.


Sent from my iPhone using Tapatalk

 

[RMerlin]

The same issue exist as with Asus's firmware in terms of port forwarding. THe simple fix for now is to simply disable HW acceleration.

Otherwise it was perfectly stable for the week that I used it as my main router (I had to revert back to the AC56U because I needed to resume development on the AC68U).

 

[nurv2600]

How is the N performance on both bands? Can I use this firmware as a reliable platform? Anything broken in port fwd or manual ips? My needs are basic, no VPN or attached USB drives, no aicloud.
Sent from my iPhone using Tapatalk

Solid as as can be. We ARE running attached drives, VPN (PPTP & OpenVPN), as well as overclocked to 1200MHz CPU, 667MHz RAM (done to improve routing and VPN performance, since we regularly have 150+ wired clients as well as multiple VPN users; overclocking was NOT necessary to attain max wifi speeds). We typically have 20-30 wireless clients between both bands, and I'm getting as much performance out of both bands as is possible: around 22MB/s 2.4GHz N (unable to test TurboQAM), 40MB/s 5GHz N, almost 80MB/s on 5GHz AC, I think it could go faster but our server can't dish out the data any faster. These speed are with the rest of the network being relatively quiet, single connection from server to wireless client, as measured by activity monitor on the client.

One thing worth mentioning that I haven't tested with Merlin's firmware was a HUGE degredation in performance copying from the LAN ports over to wireless (say, downloading a file wirelessly from our in-house server) when "Network Place (Samba) Support" was enabled in the PPTP server options (stock firmware). I can't figure out why this was, since this was all in-house, no VPN involved. It was even copying from an Mac AFP share, as well as SMB. I was getting barely 1-2MB/s; disabled that option, and suddenly the gates opened. Since I seem to be able to access our Samba servers (both on the router and our dedicated server) via VPN without it enabled, I've left it unchecked.

I can't wait to see how it all performs when Merlin gets the final version out there for our AC68Us!

 

[Dijkstra88]

Again Asus released a new firwmare update with number 3.0.0.4.374.306
Quite fast. Took about a few days from version 291.

 

[bigcid10]

The same issue exist as with Asus's firmware in terms of port forwarding. THe simple fix for now is to simply disable HW acceleration.

Otherwise it was perfectly stable for the week that I used it as my main router (I had to revert back to the AC56U because I needed to resume development on the AC68U).

on mine with your 34b1 fw portforwarding works fine with hw accel on
and loopback is working also

 

[canuckle]

Solid as as can be. We ARE running attached drives, VPN (PPTP & OpenVPN), as well as overclocked to 1200MHz CPU, 667MHz RAM (done to improve routing and VPN performance, since we regularly have 150+ wired clients as well as multiple VPN users; overclocking was NOT necessary to attain max wifi speeds). We typically have 20-30 wireless clients between both bands, and I'm getting as much performance out of both bands as is possible: around 22MB/s 2.4GHz N (unable to test TurboQAM), 40MB/s 5GHz N, almost 80MB/s on 5GHz AC, I think it could go faster but our server can't dish out the data any faster. These speed are with the rest of the network being relatively quiet, single connection from server to wireless client, as measured by activity monitor on the client.

One thing worth mentioning that I haven't tested with Merlin's firmware was a HUGE degredation in performance copying from the LAN ports over to wireless (say, downloading a file wirelessly from our in-house server) when "Network Place (Samba) Support" was enabled in the PPTP server options (stock firmware). I can't figure out why this was, since this was all in-house, no VPN involved. It was even copying from an Mac AFP share, as well as SMB. I was getting barely 1-2MB/s; disabled that option, and suddenly the gates opened. Since I seem to be able to access our Samba servers (both on the router and our dedicated server) via VPN without it enabled, I've left it unchecked.

I can't wait to see how it all performs when Merlin gets the final version out there for our AC68Us!

That, sir, is a wonderfully detailed response. THANK you! I'm picking up this router, bought a new AirPort Extreme to replace an N66U and have been underwhelmed.


Sent from my iPhone using Tapatalk

 

[canuckle]

The same issue exist as with Asus's firmware in terms of port forwarding. THe simple fix for now is to simply disable HW acceleration.

Otherwise it was perfectly stable for the week that I used it as my main router (I had to revert back to the AC56U because I needed to resume development on the AC68U).

Thanks again Merlin. I can see another donation is in order


Sent from my iPhone using Tapatalk

 

[RMerlin]

on mine with your 34b1 fw portforwarding works fine with hw accel on
and loopback is working also

I got conflicting reports on this, as other people said it was not working for them. Check the Tools -> Sysinfo page to confirm that nothing disabled HW acceleration due to incompatibility (such as QOS or Traffic Monitoring).

 

[RMerlin]

One thing worth mentioning that I haven't tested with Merlin's firmware was a HUGE degredation in performance copying from the LAN ports over to wireless (say, downloading a file wirelessly from our in-house server) when "Network Place (Samba) Support" was enabled in the PPTP server options (stock firmware). I can't figure out why this was, since this was all in-house, no VPN involved. It was even copying from an Mac AFP share, as well as SMB. I was getting barely 1-2MB/s; disabled that option, and suddenly the gates opened. Since I seem to be able to access our Samba servers (both on the router and our dedicated server) via VPN without it enabled, I've left it unchecked.

I noticed a bug in the firmware (both Asus and mine) where if you enable QoS, then wifi to USB performance is capped by the Upload speed you configured on the QoS page. If you had QoS enabled then I recommend disabling it for now.

 

[nurv2600]

That, sir, is a wonderfully detailed response. THANK you! I'm picking up this router, bought a new AirPort Extreme to replace an N66U and have been underwhelmed.

Yeah, same here at the office. We're a Mac shop, so of course, we typically use Apple hardware, but have found that they keep REMOVING features from their routers! SNMP - GONE! Visible DHCP lease tables - GONE! Static routing, multiple guest SSIDs, VPN server, web interface, QoS, - NEVER THERE! Besides, once we started getting much above 100 total clients, the whole network's performance really started to suffer.

For what it's worth Merlin, I've found no issues with port forwarding and HW acceleration enabled. We have since disabled port forwarding in preference for the built-in VPN server, to ensure security, but I do remember it working just fine, both on your firmware and the stock one; we had at least a dozen or so rules, both TCP+UDP. Hmm.

 

[nurv2600]

I noticed a bug in the firmware (both Asus and mine) where if you enable QoS, then wifi to USB performance is capped by the Upload speed you configured on the QoS page. If you had QoS enabled then I recommend disabling it for now.

Nope, no QoS (now or before). We've found between the beefy router and our business-class connection we don't need it, but may play with it later if some of our employees' Torrenting begins to get out of control

 

[JGrana]

Thanks for confirming it Paulo did a pretty good job at tracking down all those leaks and fixing them one by one.



The big limitation there is the disk IO performance - something a USB3 disk should do much better than the USB2 disk on an AC66. I would expect performance to be around 3-4x faster with the AC68U in this area.

One last (but important) point. I noticed this morning (after reverting back to the alpha4 build) that minidlna.log started to report some malloc failures. I had forgotten to enable my swap space.
So, I killall minidlna, did a swapon, and re-ran.

Ran clean as a whistle. Motto of the story - if you have large amounts of media, especially photos, you will likely need some swap space.

But, I did see that when using a Samsung TV's media client to look at the photo directories, the thumbnails appear very quickly, almost immediatly. Very nice!

 

[delancyst]

it looks like my WD 2TB Ultra cannot be detected.
Used to work on Asus original firmware before I upgraded to this.
I tried plug in and out to both USB 2, 3.0 ports but neither of the leds light up.
Rebooted the router as well.
Also tried to read the hdd off my PC and it mounts fine, so i formatted it again (ntfs), but the AC68U still can't detect it.

 

[RMerlin]

it looks like my WD 2TB Ultra cannot be detected.
Used to work on Asus original firmware before I upgraded to this.
I tried plug in and out to both USB 2, 3.0 ports but neither of the leds light up.
Rebooted the router as well.
Also tried to read the hdd off my PC and it mounts fine, so i formatted it again (ntfs), but the AC68U still can't detect it.

Look at the system log, it will give you more details as to what is happening with the disk.

 

[delancyst]

These are the only USB related log i got after rebooting the router with the hdd attached.

Curiously, theres no blue LED at the front when I try either port.
The hdd does not even attempt to power up when I plug it in.
I have 2 of these hdds, and both failed to be detected by the router.
Reverting to original firmware fixed the issue.

Jan 1 08:00:11 kernel: USB Mass Storage support registered.
Jan 1 08:00:11 kernel: ufsd: driver (8.6, Sep 25 2013 18:09:24, LBD=ON, delalloc, acl, ioctl, ugm, rwm) loaded at bf5eb000
Jan 1 08:00:11 kernel: NTFS (with native replay) support included
Jan 1 08:00:11 kernel: optimized: speed
Jan 1 08:00:11 kernel: Build_for__Asus_RT-AC56U_k2.6.36_2013-01-18_U86_r195788_b2
Jan 1 08:00:11 kernel: ehci_hcd: USB 2.0 'Enhanced' Host Controller (EHCI) Driver
Jan 1 08:00:11 kernel: ehci_hcd 0000:00:0a.1: EHCI Host Controller
Jan 1 08:00:11 kernel: ehci_hcd 0000:00:0a.1: new USB bus registered, assigned bus number 1
Jan 1 08:00:11 kernel: ehci_hcd 0000:00:0a.1: irq 111, io mem 0x18021000
Jan 1 08:00:11 kernel: ehci_hcd 0000:00:0a.1: USB 0.0 started, EHCI 1.00
Jan 1 08:00:11 kernel: hub 1-0:1.0: USB hub found
Jan 1 08:00:11 kernel: hub 1-0:1.0: 2 ports detected
Jan 1 08:00:11 kernel: ohci_hcd: USB 1.1 'Open' Host Controller (OHCI) Driver
Jan 1 08:00:11 kernel: ohci_hcd 0000:00:0a.0: OHCI Host Controller
Jan 1 08:00:11 kernel: ohci_hcd 0000:00:0a.0: new USB bus registered, assigned bus number 2
Jan 1 08:00:11 kernel: ohci_hcd 0000:00:0a.0: irq 111, io mem 0x18022000
Jan 1 08:00:11 kernel: hub 2-0:1.0: USB hub found
Jan 1 08:00:11 kernel: hub 2-0:1.0: 2 ports detected
Jan 1 08:00:11 kernel: hub 1-0:1.0: over-current change on port 1
Jan 1 08:00:11 kernel: usbcore: registered new interface driver usblp
Jan 1 08:00:11 kernel: hub 1-0:1.0: over-current change on port 2
Jan 1 08:00:11 kernel: usbcore: registered new interface driver asix
Jan 1 08:00:11 kernel: usbcore: registered new interface driver cdc_ether
Jan 1 08:00:11 kernel: usbcore: registered new interface driver net1080
Jan 1 08:00:11 kernel: usbcore: registered new interface driver rndis_host
Jan 1 08:00:11 kernel: usbcore: registered new interface driver zaurus

 

[JGrana]

These are the only USB related log i got after rebooting the router with the hdd attached.

Curiously, theres no blue LED at the front when I try either port.
The hdd does not even attempt to power up when I plug it in.
I have 2 of these hdds, and both failed to be detected by the router.
Reverting to original firmware fixed the issue.

Jan 1 08:00:11 kernel: USB Mass Storage support registered.
...
Jan 1 08:00:11 kernel: usbcore: registered new interface driver zaurus

I had the same experience with the new .374.34. I reloaded the alpha4 code and just compared the dmesg trace.

In alpha4, there is a series of xhci_hcd activity, right after the Build_for_Asus_RT-AC56U and the ehci_hcd: USB 2.0 Enhanced Host Controller driver sequence:

xhci_hcd 0000:00:0b.0: xHCI Host Controller
xhci_hcd 0000:00:0b.0: new USB bus registered, assigned bus number 1
xhci_hcd 0000:00:0b.0: irq 112, io mem 0x18023000
xhci_hcd 0000:00:0b.0: Failed to enable MSI-X
xhci_hcd 0000:00:0b.0: failed to allocate MSI entry
usb usb1: No SuperSpeed endpoint companion for config 1 interface 0 altsetting
xHCI xhci_add_endpoint called for root hub
xHCI xhci_check_bandwidth called for root hub
hub 1-0:1.0: USB hub found
hub 1-0:1.0: 1 port detected
[xhci-hub] usb2mode:[0]

then it looks like the rest of your trace. I believe xhci is related to USB 3.0 and ehci for 2.0.

Both drives are operating fine with the alpha4.

 

[RMerlin]

The xhci module isn't loaded, which means you have enabled the USB 3.0 Interference Reduction option. This option is known to cause compatibility issues at this time, re-disable that option and reboot.

My own NexStar enclosure isn't detected either if this option is enabled.