We are investigating this issue and are looking for more information to get down to the bottom of this. If you do experience this issue, please send us a private message. ... We thank you for your help and patience as we work on this issue.
So they dont know what's causing it, and they have not solved it.
Smiley Face, say it has been largely mitigated.
Ask for PMs for specific cases.
Making a widespread issue seem isolated as you implement a fix that affects 10M people.
Both iPhone users. We thought it was someone fucking with us, so I got the CNAM of the phone number and creeped the guy out by telling him his name (lol) and then we both came to the conclusion that something fishy is going on.
Tmobile with an old cyanogen build.
There are so many layers involved that it's hard to know but I'd guess it's a caching bug somewhere along the line.
Source: telecom PM for years.
It's most likely to be a bug on one of the MMSCs, but could be a problem elsewhere if the message is getting corrupted in flight. Alternatively, it could be routing issue in a certain set of numbers.
Yes MMSC errors sound right. Or wacky IMS stuff.
While I have no reason to believe that it is not the MMSC, another item to consider would be the IMS/RCS systems. I have no idea how ATT specifically has deployed their system and can't remember what vendors they are using, but there is a fairly significant behind the scenes shift towards IP based services in the wireless telco world. Personally, while I was in the industry and IMS was very new (15 months ago), I found these systems to be very immature and unreliable, so I would suspect the newer equipment before the older equipment.
Without looking into the issue in detail though, it's really hard to say. Simply speculating it could be caused by any number of problems:
1. If it's really two chat requests at a time, and two users getting crossed, I would suspect the RCS side of things, and the server actually doing the invites to the chat session could have a sort of memory corruption bug that is crossing the requests
2. If it's only a single person, then it could be happening somewhere where numbers get translated or routed. If it were say Enum though (mainly for number portability), I would expect this to affect other types of requests, like a phone call going to the wrong party, not just group chat.
3. There are a number of fairly complicated call flows, for things like interop between users who have IMS, and those who do not. But I'm not really sure how this would fail in crossover between users.
4. There is always the possibility that an Identifier somewhere being used is too short, or two vendors not agreeing on a standard somewhere
Anyway's, without digging into some of the signalling, and getting more details, it's pure speculation.
What I can say, is with IMS/VoLTE which is what I last worked on in the telco industry, we had quite a few crazy bugs.