Station Reboots with Kernel panic - not syncing: ubnt-poll-host: #8e511820

[CFWN]

I have serveral APs (mixture of RP‑5AC‑Gen2 & PS-5AC) at different locations operating in PtMP airMAX AC mode with GPS Sync enabled with Fixed 5mS DL/UL ratio of 67/33. AMDSU is disabled. A number of clients R5AC-PRISM, PBE-5AC-620 & R5AC-LITE crash at random time with the same error:  Station Reboots with Kernel panic - not syncing: ubnt-poll-host: #8e511820. Firmware is V8.5.7-CS. If I disable GPS Sync on the AP the problem goes away but this isnt a long term solution

Can anyone shed any light on this? Thanks, Mike

[UBNT-Mindas]

@CFWN - thanks for the report how frequently on average have you seen this issue? How long have you observed the device after you have turned the GPS sync off.

[CFWN]

---

@UBNT-Mindas wrote:

@CFWN- thanks for the report how frequently on average have you seen this issue? How long have you observed the device after you have turned the GPS sync off.

---

We experience the issue on average once every 24 hours - it doesnt seem dependent on load. AirView has been disabled. When we switch off GPS on the AP there have been zero crashes in over a week - hope this helps,

Mike

[UBNT-Mindas]

@CFWN - thank you, it is helpfull. Let us know if the pattern of the issue not appearing holds in say - two weeks.

Thank you.

[UBNT-sriram]

@CFWN When you switch off GPS sync, are you still in Fixed frame mode ?

[CFWN]

---

@UBNT-sriram wrote:

@CFWN When you switch off GPS sync, are you still in Fixed frame mode ?

---

@UBNT-sriram We are in Flexible new. In this mode the link has been stable for a week.

[TCILarry]

We are also seeing this very same error sporadically at different client locations around the network connected to different towers. "Kernel panic - not syncing: ubnt-poll-host: #8e511820".

I happen to have a couple of crash logs from the last client this happened to trying to figure out what the heck is causing this. They don't appear to be identical logs, but both end with the same "Kernel panic - not syncing: ubnt-poll-host: #8e511820". I'm happy to email these two logs over if UBNT would like to look at them.

I'll also include this particular AP / Station config and radio type in effort to help you guys identify a pattern.

AP is RP 5AC Gen2 XC.v8.5.8 running 20 MHz width.

-PtMP airMAX Mixed mode with GPS Sync enabled using Fixed 5ms, 67/33 ratio. This AP has AMSDU enabled, Client Isolation is enabled, Multicast Enhancement is enabled, Wireless Network Protection is enabled and ReSE is enabled.

-This particular client has a PowerBeam 5AC WA.v8.5.8 with a -63/-64 signal. 

I can't say if the others we've seen are the same type of CPE or not or the config of those devices. I hope this helps and let me know if you'd like to look at the crash logs. 

[CFWN]

Thanks for input @TCILarry. You have a very similar configuration to ours. Its a bit out there but could you confirm how the CPE is powered i.e. injector or POE switch?

[TCILarry]

Sorry for the late reply @CFWN. We power almost all of our sector radios with the EdgeSwitch 8-port POE.

[Bluestreak66]

We are seeing this one multiple stations. Some Prism Radios some not. They are all like this:

<6>[ 1692.090000] br0: neighbor ffff.00:27:22:12:c7:1e lost on port 1(ath0)
<6>[ 1692.090000] br0: topology change detected, propagating
<0>[ 1704.806000] Kernel panic - not syncing: ubnt-poll-host: #8e511820

It's always the same last 3 lines. We have had this with PowerBeams and LiteBeam 23s both gen 1 and gen 2.

We do run Fixed 5ms DL/UL ratio of 67/33 on all our APs. I've changed this to flexible on one of our links where we have a powerbeam 400 ac for both the AP and the Station. The particular powerbeam has been doing this 2-3 times a day for some time now.