R7800 data corruption when in Accesspoint mode

[pato]

Hello all
I think I discovered a bug with the R7800. I tested firmware 1.0.2.52 and now also the new 1.0.2.62. I have a NAS attached to a switch, attached to the R7800.

If the R7800 is in Accesspoint mode and I copy large files to my NAS (like 700 MB), the file will be corrupted afterwards.

This doesn't happen if it's in Router mode, nor does it happen with other access points I have for testing.

I already replaced the R7800 through warranty, but the issue still happens with the new unit.

I made enough tests now and replaced all possible parts, to be sure it's the R7800 at fault.

Anybody else could test this?

To test, copy a 700+ MB file from a wireless client to a LAN attached client and afterwards compare the MD5 sum of the two files (I suggest to use Hashtab).

I do see the TX Errors counter increasing on the ath0 (checked via telnet), but not sure if this is normal or not. Also the debug logs do show some errors.
I've attached the debug logs if you're curious.

Thanks
pato

 

[Voxel]

Hello all
I think I discovered a bug with the R7800. I tested firmware 1.0.2.52 and now also the new 1.0.2.62. I have a NAS attached to a switch, attached to the R7800.

If the R7800 is in Accesspoint mode and I copy large files to my NAS (like 700 MB), the file will be corrupted afterwards.

This doesn't happen if it's in Router mode, nor does it happen with other access points I have for testing.

I already replaced the R7800 through warranty, but the issue still happens with the new unit.

I made enough tests now and replaced all possible parts, to be sure it's the R7800 at fault.

Anybody else could test this?

To test, copy a 700+ MB file from a wireless client to a LAN attached client and afterwards compare the MD5 sum of the two files (I suggest to use Hashtab).

I do see the TX Errors counter increasing on the ath0 (checked via telnet), but not sure if this is normal or not. Also the debug logs do show some errors.
I've attached the debug logs if you're curious.

Thanks
pato

I know this problem. Replacement of the unit will not help. Workaround is to use only LAN ports in AP mode. Do not use WAN port in spite of NG manual. This fix the problem.

BTW The same problem with R7500...

Voxel.

 

[pato]

Thanks for confirmation.
Can I really use it without using the WAN port? Last time I tried that (attached the switch to a LAN instead of WAN port) my wireless clients didn't get an IP address or anything.

 

[e38BimmerFN]

Yes, when in AP mode just use one of the LAN ports. It works. Tried and tested true with my R7800...You'll be fine.

Thanks for confirmation.
Can I really use it without using the WAN port? Last time I tried that (attached the switch to a LAN instead of WAN port) my wireless clients didn't get an IP address or anything.

 

[pato]

Thanks, will try this later and update the thread

 

[pato]

Just did the first tests, and it looks indeed working correct, if the WAN port is not used
Thanks for this workaround! It would still be great if that could be fixed in the firmware, because I only discovered the problem because my backup application told me my backups were corrupt on the NAS. Otherwise I'd only have discovered it the next time I copied raw images or movies to the NAS, which would have been horrible

 

[Voxel]

ust did the first tests, and it looks indeed working correct, if the WAN port is not used

Do not worry. It is workable if to do not use WAN. One of my R7800 is in AP mode so it is tested almost two years.

t would still be great if that could be fixed in the firmware

You may try yourself to report it to NG support. Age of this bug is 4 or 5 years. Come from R7500v1.

Voxel.

 

[pato]

Thanks Voxel.
I sadly can't report it myself to NG support, because my R7800 is already > 90 days old and I can't buy a support contract in my country for this device/Netgear. But I got a PM from [name redacted] who will report it directly to Netgear

 

[XheliPilot]

I am currently using my R7800 as a wireless AP using the WAN port with no issues since FW.58 FW.60 and now with FW.63

Setup: Modem to router (A), Router (A) LAN to R7800 WAN being used as a wireless AP.

 

[pato]

Interesting. I only discovered the problem in SMB transfers and maybe SCP, never in HTTP(s). Did you also test this?
Note, my clients are all 802.11ac with at least 866 Mbps, not sure if this has any influence.

 

[Voxel]

Interesting. I only discovered the problem in SMB transfers and maybe SCP, never in HTTP(s). Did you also test this?

FYI: I faced this with SMB, FTP, SCP transfers when using WAN. Using cable, not W-Fi. So it is rather internal problem. Not Wi-Fi or specific protocol related.

Voxel.

 

[XheliPilot]

I know this problem. Replacement of the unit will not help. Workaround is to use only LAN ports in AP mode. Do not use WAN port in spite of NG manual. This fix the problem.

BTW The same problem with R7500...

Voxel.

10-4 understood.

JFYI I have always used the WAN on the R7800 as a wireless AP and have downloaded your FW and all windows updates wirelessly with no issues. Maybe I'm just getting lucky and may only occur on larger files.

I will of course change this to the LAN port per your advice. Thanks Voxel!

 

[pato]

For whatever reasons, the corruption issue is back now. The uplink is in Port 1, not WAN.
The only change I made the last few days, was to increase the channel width to 160 MHz and replaced the Wi-Fi adapter in my laptop from Intel 7260-AC to 9260-AC.
Time to test further... ARGH!

[edit]
Reboot and disabling HT160 didn't help.
One question, is it normal to have TX errors on the ath interface? I mean I would understand RX, but TX errors?
root@R7800:/# ifconfig
ath0 Link encap:Ethernet HWaddr 14:59:C0:32:BC:CC
inet6 addr: fe80::1659:c0ff:fe32:bccc/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:903280 errors:0 dropped:0 overruns:0 frame:0
TX packets:1839246 errors:480 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:945309697 (901.5 MiB) TX bytes:2705810972 (2.5 GiB)

Aren't TX errors generated when the packets already come corrupted from the CPU/Switch and then should leave the wireless interface?

 

[e38BimmerFN]

Has v68SF been tested? Its loaded on mine and it's in AP mode with the WAN port connected to Edge Router X. Not seeing any issues thus far.

 

[pato]

Never tested a third party software, only original. Anyway no new firmware since I posted this, so still open in original

 

[e38BimmerFN]

Try Voxels, it's about the same as stock FW, just better put together.

Good Luck.

 

[mr-b]

I'm running Voxel's v68SF in full router mode but configured as an Access Point and using the LAN port. That way it doesn't need to do any extra NAT so I can keep the wireless address space the same as on the LAN and use a single DHCP server for ease of client monitoring. The Netgear config wouldn't let me use the internet WAN port in the same way.

Even using the LAN interface, I still see transmit errors on the ath0/1 interfaces:

root@R7800:/$ ifconfig
ath0 Link encap:Ethernet HWaddr: xxx
inet6 addr: xxx Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:1101630 errors:0 dropped:0 overruns:0 frame:0
TX packets:2224103 errors:874 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:246774606 (235.3 MiB) TX bytes:2268827114 (2.1 GiB)


ath1 Link encap:Ethernet HWaddr xxx
inet6 addr: xxx Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:102280 errors:0 dropped:0 overruns:0 frame:0
TX packets:543661 errors:257 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:28567542 (27.2 MiB) TX bytes:177695679 (169.4 MiB)

 

[e38BimmerFN]

I'll check mine today and see what I find. Mines connected via it's WAN port for AP mode. One thing I noticed is that when using the LAN port, the Internet LED was amber, When I changed to the WAN port, it changed to WHITE. So I kept it here and it's been running like this since going on line.

 

[e38BimmerFN]

OK, checking mine in AP mode connected with the WAN:
ath0 Link encap:Ethernet HWaddr C5:9F
inet6 addr: c59f/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:6959177 errors:0 dropped:0 overruns:0 frame:0
TX packets:17845636 errors:11455 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:1040865179 (992.6 MiB) TX bytes:2885283724 (2.6 GiB)

ath1 Link encap:Ethernet HWaddr C5:9D
inet6 addr: c59d/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:19743 errors:0 dropped:0 overruns:0 frame:0
TX packets:45049 errors:26 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:3736396 (3.5 MiB) TX bytes:21301812 (20.3 MiB)

I presume this may stem from NGs side of the code?

 

[e38BimmerFN]

I was wondering about this as I haven't seen any status from this. I sent this over to NG for review. There looking into it.