RT N66U Client list bug?

[RMerlin]

Merlin

To me is not a device related issue but a general refresh issue. For the last 2 days my router is showing up 5 devices in network map. If I turn something off still shows up. If I turn something on it doesn't show up.

If I reboot the router it shows up all connected devices correctly. It seems that when you press the refresh button nothing happens, the screen on the right goes blank and the same device list appears. If you need I would be ok to do a teamviewer session so you can see what is happening....

Devices not disappearing when turned off is a known side effect of Asus's implementation. The router has no way of knowing a device is off, so as long the ARP entry exists in the cache, it will still keep showing. This is a design limitation of Asus's implementation.

Missing devices is where I need more info. The particular scenario you describe never happens to me even with all my devices being turned on one after another. That's where I need the information to determine WHY it happens to you and not to me - there is definitely something different.

 

[Deleted member 22229]

Merlin

To me is not a device related issue but a general refresh issue. For the last 2 days my router is showing up 5 devices in network map. If I turn something off still shows up. If I turn something on it doesn't show up.

If I reboot the router it shows up all connected devices correctly. It seems that when you press the refresh button nothing happens, the screen on the right goes blank and the same device list appears. If you need I would be ok to do a teamviewer session so you can see what is happening....

Could not have said it better right on target to what happens here. For me I just have a basic knowledge of routers I don't use the USB ports or any of the higher end features I just use it for routing. But clvk07 is correct I to think it has something to do with the refresh of the list. All my devices win 7 computers Roku boxes phones and tablets its all hit and miss after the refresh button is pressed. Also if I take the time and keep refreshing over and over and waiting I can get it to show the right way. On the stock firmware you click refresh and with in a second or two it shows all instantly. Thank you for your time and help. And noted about the router don't know when a device is no longer connected thats why you need to refresh and were the issue shows itself.

 

[clvk07]

To me is all devices, for example now I enabled wireless on my android phone and it doesn't show up, I turned on the second popcorn hour (a200) and doesn't show up. The first popcorn hour another a200 shows up because it was on when I rebooted the router. However the sysinfo page shows what is connected to the router correctly.

HW acceleration Enabled
Connections 703 / 300000 - 84 active
Ethernet Ports Port VLAN Link State Last Device Seen
WAN 2 1000 Full Duplex 00:17:10:01:44:8F
LAN 1 1 Down myiHome Media S
LAN 2 1 1000 Full Duplex C8:6C:87:6F:58:8A
LAN 3 1 Down <none>
LAN 4 1 Down <none>

Wireless clients (2.4 GHz) Associated: 5 - Authorized: 5 - Authenticated: 5
Wireless clients (5 GHz) Associated: 0 - Authorized: 0 - Authenticated: 0

I turned off the a200 and now the link shows down, and the wirless clients associated went from 4 to 5 once I enabled the wifi on my HTC one.

 

[RMerlin]

If multiple devices are missing, then it could be one device in particular which, when processed, prevents the processing of all other devices down the list.

The client number on the Sysinfo page is totally separate BTW. That's what is reported by the wireless driver, while the Client List relies on ARP entries.

 

[RMerlin]

Are you all using a /24 network on your LAN (meaning a netmask of 255.255.255.0)?

EDIT: I just uploaded builds of networkmap with DEBUG enabled in them (Asus left a lot of debugging code in place) on Mediafire, in the Misc directory.

First kill the running instance:

killall networkmap

Then unpack the debug build somewhere (on a USB share for example), and run it from there:

/mnt/sda1/networkmap

It will send a lot of debug output in your console as it probes your network. It might allow you to see if the scanning process gets stuck at one point.

 

[KAD]

ok so here's what I've found, which I hope will help, but I'm unfortunately not getting any log files to upload, not even the default one files in /var/

@RMerlin, you've already seen output from correct network map

so reboot using debug package via mount -o bind
network map process in gui appears to finish, however using the debug package this way I get no log files to review

only 1 device is missing from list, 192.168.1.101 which is the E3000 running DD-WRT in bridge mode


next test
disable debug file, and reboot using standard networkmap
for this test I've turned off 2 devices (laptop, and ubuntu desktop)
After reboot, we're now at 15 minutes, networkmap has not been able to finish
and as I suspected 6 devices are in list, when there should be 7
again it's the E3000 that is missing from the list

my next suspicion
I turn those 2 devices back on
refresh from gui - there should now be 9 clients, but no change
only 6 devices listed

next - again this is the standard networkmap

t@RT-AC66U:/tmp/var# killall networkmap
t@RT-AC66U:/tmp/var# networkmap

result is network is complete with no problems
all 9 clients are displayed including the E3000 bridge
so network map does work, just not at boot

my conclusion is that something is preventing networkmap from completing
and the refresh button does not kill / restart networkmap when this happens
kill and restart is only working from cmd line

and one more time I enable debug package and reboot
all devices are turned on, should be 9 clients total
but only 8 clients listed
again it's the E3000 not being displayed
simply waiting a minute it does pickup and display the E3000
so now total of 9 clients listed

so debug package does work correctly, but no log files
standard package does work correctly but not at boot, sometimes there are log files, but if it never completes log files are not created

early in this thread you mention that networkmap can take quite some time and in some cases, full list may not display until each client has renewed it's lease, it's certainly possible that this is part of what is going on

I think most people expect this to take seconds, or minutes and instead it's hours, before a lease is renewed

also it would be useful to know which file initially calls networkmap
I'd like to redirect log output, so I can reliably provide a log file

 

[RMerlin]

networkmap is launched by the rc superdaemon, so you can't easily modify it (unlike WDLXTV, things aren't being glued together by scripts, but by a C exe )

What you can do is replace networkmap with a bash script (still using mount bind), which will take care of running the real networkmap with its output redirected to a file.

 

[KAD]

ok, makes sense, should be easy to do that

I'll try and catch you late tonight with new log files

 

[cdysthe]

I am having a similar problem running the latest Merlin build. Some devices doesn't' show up with a name but rather their mac address and refresh takes a very long time.

 

[RMerlin]

I am having a similar problem running the latest Merlin build. Some devices doesn't' show up with a name but rather their mac address and refresh takes a very long time.

This isn't the same thing - the issue people are reporting here are devices not showing at all.

The delay in name resolution is normal. Networkmap tries to connect to each device, one at a time, to obtain their Netbios name. It also tries to connect to various known services (http, printer sharing, iTune) to update the client status. Once that's done, then my own code kicks in, trying to use either a device's hostname, or its description entered on the DHCP static reservation (that step takes a fraction of a second since it's just two table lookups).

I forgot the exact timing values, but the Netbios/http/etc... connection attempts can take maybe one minute per device if said device doesn't listen to Netbios requests. That's why it can be real long for names to replace MACs, especially if you have a lot of non-PC devices.

 

[clvk07]

Are you all using a /24 network on your LAN (meaning a netmask of 255.255.255.0)?

EDIT: I just uploaded builds of networkmap with DEBUG enabled in them (Asus left a lot of debugging code in place) on Mediafire, in the Misc directory.

First kill the running instance:

killall networkmap

Then unpack the debug build somewhere (on a USB share for example), and run it from there:

/mnt/sda1/networkmap

It will send a lot of debug output in your console as it probes your network. It might allow you to see if the scanning process gets stuck at one point.

Yes 24/7 255.255.255.0

I followed your steps and something strange happened.

I run the networkmap-arm and I got Invalid character so I run the networkmap-mips

It went on for a minute, and when I went back to network map and refreshed it worked. The wheel didn't hang in spinning. So I turned on another devicde and refreshed and it worked fine.

 

[RMerlin]

Yes 24/7 255.255.255.0

I followed your steps and something strange happened.

I run the networkmap-arm and I got Invalid character so I run the networkmap-mips

It went on for a minute, and when I went back to network map and refreshed it worked. The wheel didn't hang in spinning. So I turned on another devicde and refreshed and it worked fine.

the -arm build is for the RT-AC56U (ARM CPU).

So far from what preliminary result show from you and KAD is that the issue might be a timing issue. The problem doesn't appear if networkmap is run after the network is up.

Wild theory: maybe networkmap is initially run right as devices are being connected to the (still booting) router, in which case it will miss those devices that connected DURING the scan. Running networkmap manually afterward makes it start while the network is already all up and running, and properly see every devices.

That could also explain why I can't reproduce it, since my RT-AC56U has a faster CPU (so, different timing).

It's just a theory of a possible scenario that would need to be specifically tested however.

 

[clvk07]

Ye now it seems perfect I just turned on my android and clicked refresh and in 10 seconds it appeared and the wheel stop spinning. If I am not wrong, when I had the problem and rebooted the router before it was seeing all the devices so it seems that a reboot finds all

 

[clvk07]

the -arm build is for the RT-AC56U (ARM CPU).



Wild theory: maybe networkmap is initially run right as devices are being connected to the (still booting) router, in which case it will miss those devices that connected DURING the scan. Running networkmap manually afterward makes it start while the network is already all up and running, and properly see every devices.
.

forgot to ask, why with the standard asus router there is no issue?

 

[KAD]

forgot to ask, why with the standard asus router there is no issue?

we don't know yet, as the networkmap code is not changed at all

 

[KAD]

another thought, maybe it has something to do with jffs/scripts/services-start

for those of us running extra scripts, that are either reconfiguring network, or etc ...
I wonder if that has any impact

ie networkmap stars, but then some ports are reconfigured, etc ...
which causes networkmap to become unresponsive

 

[RMerlin]

forgot to ask, why with the standard asus router there is no issue?

If it's a timing issue, it could be something as apparently unrelated as I have something else that gets started before networkmap does, and that doesn't happen on the stock firmware. That means networkmap gets started at a different moment during the whole boot process.

Same reason why it happens for some people, but it never happened to me.

But until I get the chance to have this tested out, it remains a theory.

 

[clvk07]

I rebooted the router, ran the standard network map from tools cmd and I got the same issue, wheels spinning and not refreshing. Re-run the mips debug file and it worked fine again.

 

[KAD]

did you first

killall networkmap

before you tried running it again

 

[clvk07]

did you first

killall networkmap

before you tried running it again

Yes did another time the whole process same result