5Ghz goes down

[mattiL]

A radio randomly stopping functioning is most likely a hardware/design issue, and is unlikely to be resolved by a software update.

It is actually the first time this has happened to me (@5 GHz), maybe it's the start of a trend?

 

[RMerlin]

It is actually the first time this has happened to me (@5 GHz), maybe it's the start of a trend?

Or it could be just a fluke if it's the first time. Modern hardware is so complex, and operates on insanely low voltages now (compared to the TTL-based chips of 20+ years ago that ran on 5V signals), so it doesn't take much to cause something to crash.

 

[Jong]

Thanks RMerlin, but interesting it never happens with 4950, unless the hardware is just beginning to fail and interesting that sometimes it restarts and sometimes not. I agree though - wait until the replacements emerge!

 

[sfx2000]

A radio randomly stopping functioning is most likely a hardware/design issue, and is unlikely to be resolved by a software update.

Actually, it's probably a QTN software issue, and a bugger to find... been there myself back in my early days as an embedded firmware guy... everyone thought it was HW, but going deep - we found it to be a RAM memory part in an intel combo SRAM/NOR part that was same part number, but they changed FAB's, and the new memory part was a bit faster on the read side...

We fixed it by adding an additional wait-state (3 vs. 2) on the mem controller on the ARM

 

[sfx2000]

intel combo SRAM/NOR

This was almost 10 years ago - and it was a low-end mobile handset - hence the SRAM/NOR (2MB SRAM/4MB NOR flash) - we used NOR because you can execute in place (address vs. block level as NAND does), and SRAM for real time, not fast, but efficient... The "new" part was slightly faster one read, and we would read a value that was either old, or still being written, and this would thru the ARM core into error state (re-entrant data abort).

 

[RMerlin]

Actually, it's probably a QTN software issue, and a bugger to find... been there myself back in my early days as an embedded firmware guy... everyone thought it was HW, but going deep - we found it to be a RAM memory part in an intel combo SRAM/NOR part that was same part number, but they changed FAB's, and the new memory part was a bit faster on the read side...

We fixed it by adding an additional wait-state (3 vs. 2) on the mem controller on the ARM

if it were a software issue, it would happen to almost everyone, and would be fixed just by a reboot. From what I see, even after a reboot his QTN CPU fails to boot the firmware image.

 

[Jong]

I get where you are coming from, but maybe most of the time the Quantenna chip restarts and people don't notice. We don't stare at the log all the time. And there are actually a few coming forward with this now and even more talking about the normal Asus firmware who say 5Ghz "drops out". Merlin is much more chatty; I would guess this stuff would not appear in the log for the regular firmware.

The 2 times it has happened to me once the chip did restart, once it didn't. A reboot fixed it the one time it didn't.

Maybe the guy whose reboot didn't fix it did have a more serious problem, or a random corruption that needed a reset.

All in all I'd say it is too common to be a hardware fault.

 

[Jong]

No idea if this is related, but interesting that this has only happened to me since updating our devices to IOS 9: http://www.snbforums.com/threads/ac87-5ghz-ios9-weird-behaviour.25420/

 

[mattiL]

No idea if this is related, but interesting that this has only happened to me since updating our devices to IOS 9: http://www.snbforums.com/threads/ac87-5ghz-ios9-weird-behaviour.25420/

Scary stuff, same here, four devices upgraded to IOS9 just the day before!

 

[sfx2000]

No idea if this is related, but interesting that this has only happened to me since updating our devices to IOS 9: http://www.snbforums.com/threads/ac87-5ghz-ios9-weird-behaviour.25420/

Hmmm... don't think iOS9 has anything that would impact the 87U 5GHz radio - and there's been no reports of major WiFi issues with iOS9 so far...

 

[sfx2000]

if it were a software issue, it would happen to almost everyone, and would be fixed just by a reboot. From what I see, even after a reboot his QTN CPU fails to boot the firmware image.

Could be more than a single issue - as someone also mentioned earlier in the thread (Jong), the dropouts could be a watchdog event restarting the QTN. And that could be happening more often than people think...

I just find it odd that when looking at mattiL's log, the QTN is trying to get a BOOTP IP addr with a default MAC addr (QTN OUI with all zero's). So it's obviously in an odd state at that point, and that suggests bootloader issues...

 

[mattiL]

Apparently it happened again, a second time, for me today.
5GHz disabled. >==

A reboot started the 5GHz again.


As a sidenote, is there something strange with Apple IOS devices?
It seems my iPad Air 2 is requesting a new address every third minute, even when not being in active use (sleeping).

I get the following two lines approximately every third minute (filling the log file):

Sep 21 07:35:32 dnsmasq-dhcp[460]: DHCPREQUEST(br0) 192.168.1.61 2c:1f:xx:xx:xx:xx
Sep 21 07:35:32 dnsmasq-dhcp[460]: DHCPACK(br0) 192.168.1.61 2c:1f:xx:xx:xx:xx Mattis-iPad

Funny thing is, my other IOS devices with the same OS version does not do this.
Any thougths on this?

 

[tdw]

Funny thing is, my other IOS devices with the same OS version does not do this.
Any thougths on this?

Does the iPad have cellular data? And do your other iOS devices have cellular data?

 

[mattiL]

Does the iPad have cellular data? And do your other iOS devices have cellular data?

I just performed a test, switching my iPad to 2.4GHz instead of 5GHz.
No more DHCP-madness!

Must be something strange in the router with regards to 5 GHz AC network and DHCP, perhaps the reason my routes 5GHz network dies?

And to answer your question, it is an iPad with cellular capability, but I don't use it (no SIM card atm.).

 

[Jong]

As a sidenote, is there something strange with Apple IOS devices?
It seems my iPad Air 2 is requesting a new address every third minute, even when not being in active use (sleeping).

I get the following two lines approximately every third minute (filling the log file):

Sep 21 07:35:32 dnsmasq-dhcp[460]: DHCPREQUEST(br0) 192.168.1.61 2c:1f:xx:xx:xx:xx
Sep 21 07:35:32 dnsmasq-dhcp[460]: DHCPACK(br0) 192.168.1.61 2c:1f:xx:xx:xx:xx Mattis-iPad

Funny thing is, my other IOS devices with the same OS version does not do this.
Any thougths on this?

I am seeing the same thing with all my iDevices. Sometimes I just see a burst of them and then it goes quiet. Sometimes they go on and on for hours.

I'll be honest, I could not remember if this was "normal" or a new IOS9 thing. It's definitely strange.

(my iPads don't have cellular data, phones obviously do)

 

[mattiL]

I am seeing the same thing with all my iDevices. Sometimes I just see a burst of them and then it goes quiet. Sometimes they go on and on for hours.

I'll be honest, I could not remember if this was "normal" or a new IOS9 thing. It's definitely strange.

My iPhone 6 does not do this, only the iPad.

 

[mike2h]

having the the 5ghz problem on my new 87r sometimes it reboots on its own- im assuming that is whats happening with some of my 'interruptions' sometimes(like just now) I have to manually reboot. the manual reboot seems to be happening more frequently.
hopefully there is some sort of fix for this. h
as asus acknowledged the problem yet?

 

[Jong]

My iPhone 6 does not do this, only the iPad.

Less so on our iPhones, but still I see a few multiple entries. Maybe that's normal?

 

[Jong]

It's happened again this morning. Endless stream of BOOTP requests and 5Ghz network has gone down.

Sep 22 06:02:02 dnsmasq-dhcp[683]: BOOTP(br0) 00:26:86:00:00:00 no address configured
Sep 22 06:02:02 dnsmasq-dhcp[683]: BOOTP(br0) 00:26:86:00:00:00 no address configured
Sep 22 06:02:10 dnsmasq-dhcp[683]: BOOTP(br0) 00:26:86:00:00:00 no address configured
Sep 22 06:02:13 dnsmasq-dhcp[683]: BOOTP(br0) 00:26:86:00:00:00 no address configured
Sep 22 06:02:22 dnsmasq-dhcp[683]: BOOTP(br0) 00:26:86:00:00:00 no address configured
Sep 22 06:02:25 dnsmasq-dhcp[683]: BOOTP(br0) 00:26:86:00:00:00 no address configured

Interestingly, I have an AP still on ASUSWRT 4950 and it has never shown this problem (nothing in log, 5Ghz still up and running). If I can face it I might swap them over to eliminate a hardware problem.

 

[mattiL]

Less so on our iPhones, but still I see a few multiple entries. Maybe that's normal?

I'd say it is normal if it is only a few times per day, I can see in my logs that my iPad has asked for a new address only 2-3 times the last 24 hours.
It is now connected to 2.4GHz network, this seems to reduce the renewals quite a lot.