Connection errors on recent rides

Mac Mini, Ventura 13.1, and getting the black bar across the top of the screen saying ‘connection disconnected’, or some such.

Looking through the logs I’m seeing a few like:

[WARN] TCP connection timed out owing to inactivity
[INFO] TCP disconnected
[INFO] TCP host 34.217.192.177:3025 (secure)
[INFO] Connecting to TCP server…
[INFO] Saying hello to TCP server (largest wa is 1671228753093575)

Also find a few like this:

The server connection has been disconnected for xx seconds

34.217.192.177 is in the Amazon Corp domain. I’ve never seen so many (not that there are many, but from before when there weren’t any, to now, that’s a big increase.

During these messages, everything seems fine. I can see other riders, and all the functions of the trainer and HRM strap seem to work fine.

Thanks…

I am on ipad Pro and I get a message too, I think the text reads something like “a connection has disconnected” or something equally unhelpful! There’s never any indication what its trying to tell you (what has dropped? network? bluetooth to a device? which one?) but like you when the message occurs I still see other riders, my power and heartrate still display, and I can notice nothing different.

Please let me know if you do find a fix or hear back from support

The full banner across entire screen is network related an if sensors, its smaller banner in top left (well, that was the last change they made that I’m aware of).

It’s normally OK for TCP errors as the code/protocols allow for retransmission and wouldn’t normally be use for sensitive traffic (say updating fit file as opposed to rider positions which use UDP) but you two must be getting enough errors to trigger the message.

@Rob_C_Neo_2 you could try putting your log file into zwiftalizer.com and see what it says about your network traffic connections. Could be telling or a nothingburger.

@D.A you would have to get Zwift to check your errors as they have the log files for your platform.

Zwiftalizer seems to have changed, now requires an account. I’ll try, but combing through logs is like looking at the detritus under a couch cushion. You never know what you might find, and most of it isn’t usable.

Zwiftalizer kept timing out. Gave up.

The error is annoying, and apparently not fatal (yet) so I guess I’ll just ride on?

Maybe its also not coping with amount noise/errors that Zwift seem to be stuffing into the logs these days - they are certainly way larger than they used to be.

Are yuo getting it every ride?

My only concern would be that that if the errors that cause the banner start getting more frequent, it could point to a deterioration somewhere that will eventually impact your experience. Are you on Wifi - might be the whole 2.4 vs 5Ghz thing or interference from other devices (Bluetooth, microwave, fan etc). Re-booting your router never hurts either.

I feel that “a connection has disconnected” is perhaps the least useful error I’ve ever seen. It reminds of joke error messages like
image

It just tells you absolutely nothing useful or actionable. It could at least indicate what sort of disconnection it means, like “Bluetooth device XJDHDHDH has disconnected” or “Network connection to Zwift servers lost” - as it stands its meaningless/useless.

I’m not technical enough to know how to access logs on my ipad for Zwift, and really I shouldn’t have to to understand what its trying to tell me - either put enough information on the screen that indicates even roughly what the problem is, or don’t put the error at all. A bluetooth disconnect and a network disconnect are completely different things and it gives no indication at all which it thinks has occured - especially since neither has as far as I can tell!

1 Like

I did see that Apple helpfully enabled Wifi after the last update. Thinking it was that, I turned it off before doing AdZ, and saw the message something like 3 or 4 times, on direct connected gigabit ethernet and 600mb internet connection. But everything still churns along. Sometime the messages disappear in a couple seconds, and sometimes they linger a bit longer. Annoying until it gets more severe.

I’m beginning to think that the Mac Mini might not be ‘The Perfect Zwift Machine’. (sigh)

So on the 23rd I did AdZ, and had several drops. Some lasted on the screen for seconds, one I missed getting a screen cap, but others lasted for quite a bit longer. Using hard wired gigabit ethernet, no wimpy wifi here now. So any ideas? I have wiped the system several times, I have even uninstalled and reinstalled Zwift app. Just would like to not have them popup anymore.

More…

Thanks…

EDIT: This shows about 5 minutes between these two, but there is sometimes just a minute or less between them, and other times it’s quite a bit more. Didn’t know if that would matter, but… Like I’ve had one at the start of a ride, and another one towards the end. No rhythm or reason.

Around the time of the first one, I see “[16:28:27] The server connection has been disconnected for 33 seconds” in the log.txt file yet both above and below that the ‘chatter’ looks significantly normal. It looks like I’m getting communications with something, or do the 'BLEMW::Info’ lines mean it’s talking to itself? (If so, it sure seems chatty)

Prior to that I see “[16:27:54] The server connection has been disconnected for 0 seconds”, right about where it should be at t -33 seconds.

And here: [16:29:43] The server connection has been disconnected for 0 seconds
Here: [16:29:55] The server connection has been disconnected for 11 seconds
Then this block:

[16:30:00] [INFO] TCP disconnected
[16:30:00] [INFO] TCP host 0.0.0.0:3025 (secure)
[16:30:00] [INFO] Connecting to TCP server…
[16:30:00] [INFO] Saying hello to TCP server (largest wa is 1671830964656537)

And: [16:39:49] The server connection has been disconnected for 0 seconds
And: [16:40:00] The server connection has been disconnected for 11 seconds
And: [17:35:22] The server connection has been disconnected for 0 seconds
[17:35:33] The server connection has been disconnected for 11 seconds

[17:35:43] [INFO] TCP disconnected
[17:35:43] [INFO] TCP host 0.0.0.0:3025 (secure)
[17:35:43] [INFO] Connecting to TCP server…
[17:35:43] [INFO] Saying hello to TCP server (largest wa is 1671834906889736)

And then immediately after that block is: ‘[17:35:44] The server connection has been disconnected for 22 seconds’.

So it didn’t reconnect? Um…

Then: [17:36:06] The server connection has been disconnected for 44 seconds
And: [17:36:17] The server connection has been disconnected for 55 seconds

And:

[17:36:26] [WARN] TCP connection timed out owing to inactivity
[17:36:26] [INFO] TCP disconnected
[17:36:26] [INFO] TCP host 0.0.0.0:3025 (secure)
[17:36:26] [INFO] Connecting to TCP server…
[17:36:26] [INFO] Saying hello to TCP server (largest wa is 1671834951223858)

So is it connecting (seems occasionally it doesn’t)? Why is it dropping? What server is it trying to connect to, and why does the ‘game’ look like nothing happens/happened?

(0.0.0.0 – Addresses were changed to protect the guilty)

It seems more stable at the moment, rode for 150 minutes yesterday and didn’t see the message. Not sure what’s going on, but rather than being annoying (so far) this is stable. :man_shrugging:t2:

Another two, at least, on the 31st.

Is this a server load issue, or is it my ISP or connections?

I have the same issue. Computer is connected to the modem with an ethernet cable so it’s not a wifi issue. I’m also on macOS Ventura but I only upgraded a few days ago. I had the error message with Monterey too. I first noticed the message around the holiday period.

[10:36:45] Starting critical power curve search job (244493352)  mTimeStamp = 1494.50964  m_lastSearchTime = 1484.50354   calculating=false
[10:36:49] The server connection has been disconnected for 11 seconds
[10:36:50] [WARN] TCP connection timed out owing to inactivity
[10:36:50] [WARN] Error shutting down TCP socket [49] Can't assign requested address
[10:36:50] [INFO] TCP disconnected
[10:36:50] [INFO] TCP host 34.210.80.71:3025 (secure)
[10:36:50] [INFO] Connecting to TCP server...
[10:36:50] [INFO] Saying hello to TCP server (largest wa is 1672673773025263)

I had three or four yesterday. If it’s benign, why is it being reported to the user. It’s a nuisance message. What’s amazing is that the app, while it’s running, is very chatty, but apparently not talking to the server using TCP. So why is the countdown timer so short, and why report the ‘error’ when it’s apparently nothing. Who knows. No one who might know has responded.

Is there another thread somewhere about this issue?

And glad to see I’m not the only one…

AND the IP address 34.210.80.71 is an Amazon Technologies address. (I thought it might be mine which is why I edited it out in my posts) So it’s Amazon that is causing this. Odd. They should be able to handle the load, but it could be load based as Amazon only supports the traffic estimates you pay for. Hmm…

What’s your ISP? Spectrum/Charter here.

Any chance you are running out of CPU? That could cause a timeout to occur on the client side…

I’m running a 3.0GHz (turbo boost to 4.16GHz) 6-core Intel Core i5 Mac Mini. (Gigabit ethernet, 600m internet) Is it possible? Anything is possible, but I would tend to doubt it. Plus this hasn’t been a long time issue. I think this started on the last update, or possibly the one before. I was thinking it could be on Amazon’s end. They aren’t unknown for having some configuration issues on their servers occasionally.

I could run Top in the background, but it has an effect on CPU load too. I can also try dumping the network traffic from the Mini and see what it’s doing, but that won’t necessarily be very instructive. I can’t find any information on what Zwift is using 3025 for to know what the impacts are, and if it really is a timeout issue and if they need to tweak that to stop the notifications. I found conversations where Zwift programmers are using successive port numbers for functions in game, but nothing (so far) showing what port 3025 s being used for.

Maybe I should block it and see what’s broken. Apple lists what ports they use, and they are notorious for being locked down/paranoid. What is 3025 used for, and why is it timing out?

I could try the M1 backup mini, although I don’t know what that would prove if the messages stopped.

Whether it’s affecting you here or not I can’t say, but CPU load is higher than it should be on the last two updates and there’s some weird memory utilisation issues going on.

I just completed a ride, I had a bunch of disconnects again. In my case the other riders disappear for a few second, come back and then the error is displayed. On a few of the disconnect the companion app also goes back to the main screen and then maybe a minute later return to the map screen.

I wonder if using a proxy app like Charles could provide more information on the connection drops. I will try it.

I’m in Canada, my ISP is Vidéotron.

I have a similar config to Robert, I5 2.8GHZ (6 core). Plenty of power for Zwift. But you make me think, it could be the CPU on the router. Depending on what kind of latency Zwift support it may cause a connection to drop if the router struggle with the load.This is just a guess, I have no indication this is happening…