Catalyst 9800-CL client losing connectivity

[Jeff A.]

I've got a strange issue which I'm not quite able to find the root cause for.

There is one particular client type on my network (iPad) which will, after some time, loose network connectivity as well as its IPv6 address. The affected client usually loses connectivity after being on the network for a few days. Once connectivity is lost, nothing on the network can be pinged, either on IPv4 or IPv6. I don't know if the IPv6 address is just the first one to fall or not as the IPv4 address is sticky. Performing a DHCP lease renew does not bring the client back online, instead the client must be disconnected from the wireless network and re-connected before restoring connectivity.

I have a dual-stack LAN where the IPv6 addresses are provided to clients using an external router via RA. AP's are in FlexConnect mode with local switching but central authentication. The issue appears to re-occur every 3-5 days of the device being connected to the network. Since the device does not leave the premises, it will eventually fail. Either other devices do not appear to be affected (Macs and iPhones) or there is some timeout that is occurring at the time when the affected clients are in use. Next time it re-occurs I will attempt to find out how long, if at all, it would take for the client to have its connectivity restored.

I was initially experiencing this as a more widespread issue across iPhone devices where even the IPv4 address was lost, and per the advice of another discussion thread on this forum I modified the Session Timeout on the WLAN to 0 seconds, which appears to have resolved that. Now it just appears to affect iPad devices, but I'm not sure why they would be losing connectivity and IPv6 addressing but not IPv4 addressing.

Any insight or suggestions would be appreciated.

[Scott Fella]

Just try to eliminate variables. Maybe it’s ipv6, does this happen on Ethernet, does this happen if you bring the iPad home or switch to a different network from time to time. I have an iPad Pro I use and never had any issue with it. I run a setup at home with 6 access point in FlexConnect. The only issue I have is with iOS 14, the devices prompt to enter the password a few time a day. So try to isolate the issue by seeing what works and doesn’t work with these iPads you have. Maybe it’s best to have them restart every night. Have you posted in the Apple community forums?

[Jeff A.]

It's funny you say that - I had issues with IPv6 RA from my ISP which is what I thought was causing these issues. Since working with the ISP and fixing that mess up in my router, I'm only having this last issue with the iPads, otherwise everything is smooth sailing. (except for one Sonos speaker having issues with AirPlay, but Sonos are renowned for their instability on enterprise networks so I'm not bothering to troubleshoot that further)

[Scott Fella]

Set session timer to something other than 0 like grendizer mentioned. This fixed my issue but that is different from what you are seeing but will probably prevent other issues.

[hd_nw_net]

Hi,

what do you mean with:

1) Disable FT completely from that WLAN, FT will not save that much of time if the WLAN is PSK.

what is FT?

[Jeff A.]

FT is 802.11r BSS Fast Transition.

[Jeff A.]

Since disabling FT completely, including adaptive mode, and ensuring the session timeout was not 0 but instead set to 86400, the issues have not re-occurred. Thank you to all who helped and provided advice.

I am curious to understand more as to why these two items caused the issues experienced. Does anyone have any resources to help me dig deeper into those aspects?

[Scott Fella]

Well the best thing is to try disabling one at a time. The session timer is a bug which was mentioned in the thread. So maybe if you want FT, enable that feature and see. Bugs will be issues with the wireless, but you have to also understand that device manufacturers might implement other features differently and possibly not work or work well.
What I have noticed since I set session timer to 86400 is that I’m no longer seeing my Apple devices asking for my password. However, my iPhone at times seem to not have network or is slow and I still have to toggle the wireless off and on.

[Jeff A.]

Thanks Scott. Funny I just noticed that a HomePod dropped offline in that manner. The end user devices have been solid so that's the priority, but I'll probably downgrade the controller version, maybe even go back to the starred version, to see if that helps with that overall stability.

[Scott Fella]

Well keep us posted.

[Jeff A.]

Downgrading to 16.12 seems to be a lot better for my environment, although I am missing AVC in the IPv6 space now.

Anyway, stability is more important than metrics which I rarely look at or make any difference to my users at the end of the day.

[Scott Fella]

Well good for you. Metrics are nice, but stability is key. I don’t look at metrics or care for features is everything is working well and users are enjoying their experience.

[Jeff A.]

The other benefit is being able to use 0 as the session timeout rather than 86400. The latter worked fine for iPads and iPhones, but was kicking Apple TVs and HomePods off the network every few days.

I did some tests with FT as well and the difference between 802.11i and 802.11r with PSK was 5-10ms, doesn’t seem worth the drama since all my 802.11i roams were occurring in around ~18ms anyway.

[Scott Fella]

I should do more testing myself but would have to do it after my kids are done with school or else I will never hear the end of it. I do notice that at times even though I’m connected, seems like data stops passing. I would have to disable and then enable wireless, but only notice that on my iPhone.

[Jeff A.]

Interesting, that sounds awfully similar to my original problem. Is your timeout set to 86400 and FT turned off as well?

[Grendizer]

Thanks for the update Jeff...