No job cleanup (or resume) on startup

The ongoing recording/comskip jobs active during the shutdown are still shown under activity. They should be resumed/restarted or cleaned up.

what kind of shutdown are you referring to?

steps to reproduce and an example of what you're seeing would be useful.

via systemctl

Perhaps your browser was showing cached information then, because Activity would be blank after a process restart.

Nope. The /dvr endpoint shows the same.

what does it show?

The only thing shown there is current activity. So if you're seeing things in there it means they're running. They resumed on restart.

You are right! That's very nice. However live sessions get interrupted.

That's because when the server restarts the clients cannot access the stream, so the client app closes the connection. In that case the onus is on the client app to re-establish the connection.

Do note, if you are using a HDHomeRun tuner with the default settings—tuner sharing being disabled—your live stream will continue uninterrupted, because live HDHR streams to the clients completely bypass the DVR server by default.

I don't see the related ffmpeg processes on the server though for the restarted jobs. I thought they were important :wink:

channels-dvr/2021.12.01.2223/ffprobe -hide_banner -print_format compact -show_packets -

That process is involved in indexing recordings while they are being recorded.

Any recording that was interrupted during recording can no longer be indexed while it's being recorded and will be indexed once the recording has completed.


I think there is a problem with service worker as the request for manifest.json is made without the session cookie so it fails when connecting remotely....

Not an option for a remote connection :wink:

Works for me.

In a couple of hours everything goes back to normal ...

I think I've been seeing something similar for a while. Every time I update Channels DVR via Settings, I end up in a reload loop in both Safari and Chrome on multiple machines where it rapidly alternates between a small reload icon at the top and a bigger one in the middle of the screen. It's hard to see what's going on in the browser console because it's reloading and clearly so quickly, but I do see an error related to "manifest.json" before it clears and reloads again. I've tried clearly the cache, local storage, cookies, and only time seems to fix it. As mentioned above, I'll come back in a few hours or a day and the Channels page loads just fine. The pattern repeats every time I update, and it's been happening for at least a month or so. All I see the logs is from channels-dvr-http.log and it just repeats this pattern over and over:

2021/12/18 12:38:37.549169 [HTTP] | 200 |      75.181µs | | GET      "/manifest.json"
2021/12/18 12:38:37.645904 [HTTP] | 200 |     1.13234ms | | GET      "/status"
2021/12/18 12:38:37.665447 [HTTP] | 200 |     310.768µs | | GET      "/admin/"
2021/12/18 12:38:38.053591 [HTTP] | 200 |     713.024µs | | GET      "/auth"
2021/12/18 12:38:38.132863 [HTTP] | 200 |   11.063785ms | | GET      "/status"
2021/12/18 12:38:38.161506 [HTTP] | 200 |      74.755µs | | GET      "/manifest.json"
2021/12/18 12:38:38.331777 [HTTP] | 200 |     330.598µs | | GET      "/admin/"
2021/12/18 12:38:38.656688 [HTTP] | 200 |     712.255µs | | GET      "/auth"
2021/12/18 12:38:38.729931 [HTTP] | 200 |    1.013473ms | | GET      "/status"
2021/12/18 12:38:38.732325 [HTTP] | 200 |      64.102µs | | GET      "/manifest.json"
2021/12/18 12:38:38.900510 [HTTP] | 200 |     341.074µs | | GET      "/admin/"
2021/12/18 12:38:39.230649 [HTTP] | 200 |     755.292µs | | GET      "/auth"
2021/12/18 12:38:39.294583 [HTTP] | 200 |     994.865µs | | GET      "/status"
2021/12/18 12:38:39.355460 [HTTP] | 200 |      80.453µs | | GET      "/manifest.json"

Hold shift and click the browser refresh button

I've done that often and it doesn't help. I've also opened the developer tools panel, right-clicked on refresh, and selected "Empty cache and hard reload". I've also tried disabling the cache in the dev tools. I've used as big a hammer as I can find, nothing seems to fix it, only time. I have no doubt that it's some sort of cache and timeout, but the same thing happens across browsers and machines and I haven't found any way to clear out whatever is lingering. I've cleared "local data", "cookies", everything I can find in the "storage" pane, without success.

It was happening last night, and as expected, it's working this morning.

Edit: Following up, I was accessing channels via my NGINX instance and it looks like disabling "Cache assets" in NPM seemed to help. So perhaps it wasn't my browser as much as NGINX.