Closed stigger closed 1 month ago
Looks like a great catch. Is there any specific device this has been verified on? I remember testing the suspend states on the SHIELD Pro 2019 and it worked flawlessly.
I don't doubt this fix, I'd just like to know if this is device specific :)
I encountered the problem on Ugoos AM6B+: always reproduces when the hibernation time was longer than 5 minutes (the UpdateHubTask
period). Works correctly for me after the change.
Interesting. This PR is fine, but I think it proxy-fixes a kodi/elec bug to be honest. The network device should completely be there once the python subsystem gets reactivated via the event.
I'll merge this commit, as it doesn't negatively impact anyone else, but could you make sure this issue gets reported back to the kodi/elec maintainers for this device?
Thanks!
The problem is not that the network is unavailable, the problem is that the client doesn't know that the server has closed the connection for some reason (e.g. tried to send a keepalive to the client while it was hibernated). When the client wakes up and tries to use the previously established connection, this results in an error, since the connection is not valid on the server side anymore.
I don't think there is anything that should be fixed in Kodi or CoreELEC. Although I can only guess why this does not reproduce for you as easily, I'm sure that in certain situations the issue can be triggered even for you. The easiest way that should do the trick: hibernate the Shield, wait 5 minutes, restart your Plex server, wake up the Shield.
Sounds reasonable to me. Thank you!
Description:
After restoring from the hibernated state, all previously established TCP connections are being reset. As a result, the section refresh triggered by the
system.wakeup
listener fails, so the user have to switch to some other section in order to see any content. Allowing retries fixes this problem.Checklist: