Closed pogzie closed 3 years ago
This is something I've had happen a few times, and it is really, really tricky.
So basically, if you get a network hiccup, you've got a few things that can happen:
The RTMP module closing the connection causes the most frustration. When that happens and OBS eventually reconnects, the follow things happen:
I think (though I'm not entirely sure) that this can be controlled a bit with the ping
and ping_timeout
directives, here: https://github.com/arut/nginx-rtmp-module/wiki/Directives#ping - but the behavior can still get pretty weird.
I'll have to experiment some more. It seems like shorter outages (<= 30 seconds) are fine now that ffmpeg will try restarting a few times.
OBS was able to recover, but none of the stream was unable. Let me know if you need more logs or info.
Yeah, I'm not sure what to do on this issue. I created short network hiccups (~10 seconds) and everything recovered fine. Any longer, it seems like no matter what combination of timeout, ping_timeout, etc -- either OBS never succesfully reconnects, or the RTMP module winds up calling "on-done" followed by "on-publish", which ends existing videos and starts new ones. I don't have a good way to detect a dropped connection vs somebody actually clicking "end stream" in OBS.
There's a bit of a work-around on the Facebook module - you can choose a pre-existing live video and uncheck "End live video when stream ends" - I wrote that specifically for a time I did a 24-hour marathon stream on Facebook, so I could disconnect/reconnect.
Implementing similar functionality in the YouTube module wouldn't be too hard. But for both the Facebook and YouTube modules, you'd have to create the event first on Facebook/YouTube, then log into Facebook/YouTube when you want the video to end instead of just ending the stream in OBS.
not sure if this helps you, but the issue can't be reproduced with Xsplit
This happened again when my network connection got knocked out. My PC (gaming, OBS) is separate from my Multistreamer server. I had to stop/start the stream from OBS.
I think the only info I got was I waited for it if it would recover but after a minute i decided to just restart the OBS stream.
I've been considering discontinuing this project for some time now. I'm sorry I couldn't solve your issue, but I just don't enjoy working on multistreamer anymore.
I was getting timestamps for more information on the start/end discrepancies between Multistreamer and all the services and halfway through playing/streaming my OBS just gave out an alert that I was disconnected and its trying to reconnect.
All the while I thought that its just gonna be a hiccup and everything would continue on as it should. As I completed my stream, I noticed that the OBS disconnect cut the stream altogether. Is it by design? Or should Multistreamer be at least able to recover on things like this like a network hiccup or some quick OBS failure?
Previously in one service im using, even if OBS disconnects, there would just be a gap on the stream but would still continue on when OBS resumes the stream/reconnects.
To help out further, here are the start timestamps of the videos and services.
YT: Started 23:51:31 (lasted for 00:15:38) FB: Started 23:51:14 (lasted for 00:16:02) Twitch: Started 23:51:30 (lasted for 00:15:46)
Please note that these timestamps are GMT +8. I have gathered the logs but unforunately its in UTC.
I wish I could do the math but I would need to sleep. So the logs in question should be around 15:51 UTC at the very least.
20180319 Multistreamer Logs.zip