Sunday, 22 January 2017

UBNT/Ubiquiti and iPhone and roaming not working

I have posted a few times on this - the issue is plain...

If I go to the toilet my iPhone stops working. Obviously everyone takes their phone to the toilet, don't they. It is not just me. Tell me it is not just me?

If I sit in the bath in the morning trying to catch up on twitter and FaceBook my iPhone stops working.

This is all about roaming between Access Points on the WiFi, when I roam between APs (I have 3 in the house) I lose connectivity. It is annoying.

This has varied from occasionally to damned annoying, depending on s/w versions. The latest code on the iPhone and Ubiquiti led me to turn off WiFi and read Twitter in the bath on 3G as I kept having to reset the WiFi. That really is the last straw.

Ubiquiti are now talking to me and working on it. Thank you.

We have tried many things, but the last few days I tried turning off the IPv6 on my LAN. Yes, that was strange. Most of what I do uses IPv6. Most of it will fall back to IPv4 (but not all). I have machines (e.g. my main Mac) with fixed IPv6 addresses but router set via RA, so that caused delays on access to all sorts of things. Turning off IPv6 is strange. It is, after all, the current version of IP. I'd be happier turning off IPv4. I have turned it back on now. Phew!

Initially it did not fix anything, but the iPhone still had IPv6 addresses even though RA was off. Looks like my RA code sent "infinite" life on prefixes on removing IPv6 RAs rather than zero - oops. Fixed. But proved RA itself is not the issue, it is the iPhone having IPv6 addresses.

Once it had only IPv4 addresses it worked, and roamed between APs seamlessly with no problems for 2 days! Turned IPv6 back on and the problem came back immediately.

So, is it an iPhone issue or a Ubiquiti issue? Hard to say. I do not know if "roaming between APs" has any "is this IP OK still" logic? Or what? I don't know enough of the protocols.

Many people have said they use Ubiquiti and see no issues. Ubiquiti have said a few people have issues, and they don't understand why. This could be it.

The good news is Ubiquiti are working on it - trying to reproduce the issue. I have offered a FireBrick that will send RAs for them to test. Let us hope this is close to the end of the issue. If it is iPhone, I hope they will raise with Apple.

If we can only solve this, the Ubiquiti are excellent APs at a sensible price.

6 comments:

  1. A client (makoto can confirm) has constant problems that sound very much as you describe. They have four UniFi APs on a FB2700 and as they're a media company there are lots of iProducts. Windows/Android users there have very few issues.

    I'm looking forward to how this works out.

    ReplyDelete
  2. Which firmware are you using? iPhone / Apple issues have especially been fixed in recent controllers + firmware, make sure you are running controller 5.4.9 and firmware 3.7.37, which appears to be very reliable.

    ReplyDelete
  3. https://cdn.meme.am/cache/instances/folder388/29178388.jpg

    ReplyDelete
  4. There's nothing inherent in WiFi that involves IP as far as I know.

    Most devices (though interestingly enough *not* Android) for IPv4 will send a DHCP REQUEST packet upon roaming, and if they get a NAK back will do a full DISCOVER cycle - this is useful as it handles the case where you roam between APs that are the same SSID but not the same subnet (e.g. something that is quite widely available like eduroam).

    It's possible there is something similar for IPv6 that is breaking in the case of Apple - have you tried capturing the packets from two APs while roaming occurs to see if there is anything interesting there?

    ReplyDelete
  5. Unfortunately my company has a policy of not reporting bugs to Apple these days - each time we reported a bug they asked us to spend a lot of time collecting debugging data and then dropped the bug report in the bitbucked. We haven't had a single bug fixed by Apple so we concluded that reporting them was just a waste of our time and now don't. (I don't mind spending time collecting debugging data if they are actually going to use it, but "collect all this data" seemed to be their default response, even if we had included it all in a slightly different form in the original report, and they ignored the bugs after we'd done all that work).

    ReplyDelete
  6. Anecdotally I think I experience this, though I've never really tried to diagnose it.

    At work (FB2700 and 3 x Unifi APs) various people on Apple devices complain that the wifi is "unreliable" and at home (RouterOS and 5 x Unifi APs) I have the exact same issue you do - crappy connectivity on my iPhone when I am, ahem, indisposed, which is fixed by disabling and re-enabling wifi on the device.

    I don't have any IPv6 at home though.

    I'll check firmware

    ReplyDelete