Why isn’t there a forum section to talk about Pixel Streaming? I am only a 3D modeler. I have no idea of networks, IPs, ports, cloud servers… I can not react when an error message appears related to those areas. I am managing to stream after many weeks looking for information (practical cases) beyond from Epic docs.
I also observed when it worked the first time, that my friends were connecting from their phones with no problem, but when any one connected with “huawei” phone, SignallizingServer stopped working. We repeat the test with those models a few times. No problems with the rest of the smartphones.
Here you can see the log when it’s happens:
18:29:49.863 <- player 101: iceCandidate
18:29:49.933 <- Streamer: {“type”:“disconnectPlayer”,“playerId”:101,“reason”:“Failed to set remote offer sdp: Failed to set remote video description send parameters.”}
C:\Users\DANZZA\Desktop\streaming\WindowsNoEditor\Engine\Source\Programs\PixelStreaming\WebServers\SignallingWebServer
ode_modules\ws\lib\sender.js:108
throw new TypeError(‘First argument must be a valid error code number’);
^
TypeError: First argument must be a valid error code number
at Sender.close (C:\Users\DANZZA\Desktop\streaming\WindowsNoEditor\Engine\Source\Programs\PixelStreaming\WebServers\SignallingWebServer
ode_modules\ws\lib\sender.js:108:13)
at WebSocket.close (C:\Users\DANZZA\Desktop\streaming\WindowsNoEditor\Engine\Source\Programs\PixelStreaming\WebServers\SignallingWebServer
ode_modules\ws\lib\websocket.js:228:18)
at WebSocket.onStreamerMessage (C:\Users\DANZZA\Desktop\streaming\WindowsNoEditor\Engine\Source\Programs\PixelStreaming\WebServers\SignallingWebServer\cirrus.js:295:14)
at WebSocket.emit (events.js:311:20)
at Receiver.receiverOnMessage (C:\Users\DANZZA\Desktop\streaming\WindowsNoEditor\Engine\Source\Programs\PixelStreaming\WebServers\SignallingWebServer
ode_modules\ws\lib\websocket.js:800:20)
at Receiver.emit (events.js:311:20)
at Receiver.dataMessage (C:\Users\DANZZA\Desktop\streaming\WindowsNoEditor\Engine\Source\Programs\PixelStreaming\WebServers\SignallingWebServer
ode_modules\ws\lib\receiver.js:422:14)
at Receiver.getData (C:\Users\DANZZA\Desktop\streaming\WindowsNoEditor\Engine\Source\Programs\PixelStreaming\WebServers\SignallingWebServer
ode_modules\ws\lib\receiver.js:352:17)
at Receiver.startLoop (C:\Users\DANZZA\Desktop\streaming\WindowsNoEditor\Engine\Source\Programs\PixelStreaming\WebServers\SignallingWebServer
ode_modules\ws\lib\receiver.js:138:22)
at Receiver._write (C:\Users\DANZZA\Desktop\streaming\WindowsNoEditor\Engine\Source\Programs\PixelStreaming\WebServers\SignallingWebServer
ode_modules\ws\lib\receiver.js:74:10)
Press a key to continue . . .
I got the same issue, when I reconnect through firefox after about one hour, I got the message below:
15:45:04.046 <- player 102: iceCandidate
streamer disconnected: 1006 -
15:45:22.950 player 102 connection closed: 1005 -
Same issue, same error on 4.24, been successfully using PixelStreaming on Azure with 4.22 builds, and could connect dozens of devices without any crashes for days. Now, with 4.24 it works one time only, meaning whenever same device connects again, or different device connects as a second connection, powershell script throws errors mentioned above, and you have to re-run the turn bat.
I’m gonna try with 4.25 and report back.
Ninja Edit:
It seems that apart from more streamlined approach to setting up whole server/webrtc logic, user disconnection has also been changed or rather overhauled. Cirrus.js file is no where the same as pre 4.24, and since I cannot tell exactly what’s happening in there (lack of js experience), I’m stuck. Still trying 4.25 on my TODO list.
Done 4.25 p7 build and same problem occurs with not connecting users, although cmd doesn’t crash anymore it is still unable to connect users after few connection attempts.
I have the same problem. It does not work at all on mobile devices, it crashes when a second device is connected. An urgent solution is needed, or at least a temporary one!
That is correct. Apparently changes to the plugin made in 4.24 have made Pixel Streaming unreliable. What I managed to observe was that randomly either npm and/or app running offscreen crashed, although logs does not show any critical/crash errors, just logs that it closes the application since a function to close the app been called (why though, no idea).
All above refer to 4.25 p7, since on 4.24 npm crashes and prompts to close the window, with 4.25 server is running even though it somewhat crashed, so you can still access player.html by going to public ip setup for pixel streaming.
To me it seems like there’s something weird happening under the hood as far as pixel streaming plugin goes. The whole cirrus.js/*.bat files setup and config is only streamlined to a point where you need to run one *.bat and an app shortcut versus four *.bats and a shortcut in 4.23 and lower.
Hey all, for me the streamer just randomly disconnects after few minutes. The app is still running without error. It seems like it just stops communicating after a certain amount of time.
Could it be a timeout? Anyone know if we can reconnect to cirrus through blueprints or something?
It seems it depends on the browser and its version. In my case it only works with Chrome 88.0.4324. With Chrome 89 it crashes with “…player connection closed”! Other browsers like the current Firefox or edge do not work too.
Assertion failed: Player [File:D:/Build/++UE4/Sync/Engine/Plugins/Media/PixelStreaming/Source/PixelStreaming/Private/Streamer.cpp] [Line: 173] player 102 not found
Also, I noted that the screenshots on the Pixel Streaming documentation on getting started are not consistent with the actual steps that must be followed. The differences are minor (such as referring to a power shell script that doesn’t exist) though it was possible to figure out what was intended after examining the files that were present.