Re: Terrible Bug - Adding a RBS20 satellite to RBK 50 system breaks the LAN

superczar · ‎2019-10-21

Think I figured out the issue (though it may be too early to say for sure)

As strange as this may sound, The Fast-roaming flag does seem to be the underlying issue.

The symptom was lack of access to devices connected to the satellite 1 to LAN devices on router/sat2.

After having tested this out for a day, it appears that the problem does not occur if fast roaming is disabled

This would lead me to believe that the MAC lookups (ARP?) for devices that switched from router to SAT1 are getting messed up somehow if FR is enabled - With FR disabled, the Orbi (Probably ) works like a regular multi Access Point LAN and the issue doesn't happen

The new FW release note seems to have fixes for Iot devices so maybe that fixes the LAN issues - will try and report

Alfiegerner · ‎2019-10-21

Hi,

Thanks for all your investigation on this, you've jumped through a bunch of hoops I wouldn't have thought of. - I have the same set up (RBR50 + RBS50 + RBS20) and have also have horrible ping times on local lan.

I've updated to latest firmware, V2.5.0.38 for RBR + S 50, and V2.3.5.26 for RBS20, and seeing same results, latency can look alright for a short while before spikes of > 1000ms come in, then persist.

As the issue can be sporadic I'm not 100% sure if the RBS20 satellite is the key, particular as there are other user who seem to have similar issues without additional satellites.\

I'll keep on tinkering and let you know if I find anything

Alex

superczar · ‎2019-10-22

In my case, I lose connectivity altogether between some devices on RBR50 and some devices on the rest of the network . e.g. A Macbook (device A) , when connected to RBS50 cannot reach let's say my Hue unit (device B) which is connected wired to the RBR50

One definitive that I have noticed is that switching off the RBS20 resolves all the issues almost immediately

One likely link I have noticed over the last two days is when I switch off fast roaming, the network has been running stable.

Tried this 3-4 times now - When I Enable fast roaming , device A can reach dev B for some time , 10-15 mins later, A cannot reach B at all

Then I disable fast roaming and reboot - Device A can now reach device B even after say 10 hours.

The last FR toggle I did was 12AM local time (Indian Standard time). It's 2 PM now and the network is stable.

superczar · ‎2019-10-23

I called up support - They have taken down the case details and have forwarded it to L2 support.

Let's see if something comes out of it.

On a side note, it's not been more than 24 hours and everything is running stable (with Fast Roaming Off)

Whatever the issue it, It is most certainly related to the Fast Roaming flag

FURRYe38 · ‎2019-10-23

Nice find. Would be interested to know if NG fixes this for you and what results are.

superczar · ‎2019-10-24

The problem started again after running stable for a day

Enabling FR breaks LAN-LAN communication between satellites almost immediately

Disabling and rebooting gets it working but only for some time

This is most certainly not an isolated issue and many others have reported it on the RBS20

https://community.netgear.com/t5/Orbi/Orbi-RBK23-LAN-Connections-Fail-Cross-Satelites/td-p/1693135

A bad LAN should have been a very high priority bug fix for Netgear but I guess they just aren't bothered as LAN-LAN is not very important for majority of users

FURRYe38 · ‎2019-10-25

Something to submit a support ticket for and submit your debug logs too.

Something that some users haven't seen however don't have this same configuration. Possible there is a problem with the RBS20 being introduced.

superczar · ‎2019-10-27

Haven’t heard back from L2 support as yet but it's holiday season in India at the moment.

In the interim, I downgraded RBS20 to 2.0.0.64 which made things a little more stable but not completely.

With the assumption that the issue has something to do with a bug pertaining to STP, i have borrowed a dumb gigabit switch and swapped it out for my SG108e.

After putting that in, it appears that (at least for now), all devices on all satellites/wired/router can see each other.

Will post back in a few days to see if this runs stable

superczar · ‎2019-10-29

Couple of more updates and findings.

The network broke again because the satellite auto upgraded itself (even though I had blocked all services to the RBS20 IP address)

So did it the hard way - Reconfigured everything to go back to old topology of Orbi as AP and Edgerouter - X as router

Blocked Internet access to all 3 Orbi devices (so that they can't do this silly auto update business)

Downgraded all 3 to 2.1.4.16

Seems to be stable for now - Only time will tell

PS: One odd finding:

I decided to run a program called arpwatch on a connected linux machine to see if i can find something that's causing this issue

Am not sure if it's related but the RBS50 keeps sending a bogus ARP flood of 192.168.1.250 every few seconds

Oct 29 18:14:53 mint arpwatch: bogon 192.168.1.250 8c:3b:xx:xx:xx:xx ens18
Oct 29 18:14:53 mint arpwatch: message repeated 20 times: [ bogon 192.168.1.250

(The MAC above is of the RBS50 which has a static IP of 192.168.5.13)

This is also referenced here - https://community.netgear.com/t5/Orbi/RBR50-amp-RBS50s-disconnect-from-Android-phones-amp-tablets/m-...

My nework is on 192.168.5.0/24 so the RBS20 has no business doing this

On a related note, can't feel but amazed as to why would Netgear use such a ridiculously bad firmware and coding practices on hardware that has so much potential

superczar · ‎2019-10-31

Just an update that everything is stable and working fine now .

Orbi is in AP mode and all Orbis are on 2.1.4.16

FURRYe38 · ‎2019-10-31

Sounds like some form on incompatibility in recent FW between the RBS20 and the RBK50 system. Maybe NG can look into this.

@Blanca_O

@superczar wrote:
Just an update that everything is stable and working fine now .
Orbi is in AP mode and all Orbis are on 2.1.4.16