Cleanup never matched a live session because the controller registered empty
manifest and segment paths, renewal was a read-modify-write that could revert a
takeover, and the takeover script compared an ISO date to a number, so it errored.
- populate the session record's manifest and segment paths from the playlist path
- renew the lease via a Lua compare-and-set on the owning pod
- store the lease expiry as unix milliseconds so the scripts can compare it
- retain the session record past its lease so an orphan can still be taken over
- test the Redis store against a real Redis, including the renew-vs-takeover race
- drop the live stream record nothing ever read back
A transcode that is not in the store is invisible to the other replicas, so the
HLS entry points have to publish and hold the lease themselves.
- register the play session in the store when ffmpeg starts
- renew the lease on a background loop and delete the session when it ends
- shorten segments and bound the playlist window when resuming a stored session
- add RecoverySegmentLengthSeconds and RecoverySegmentBufferCount encoding options
This is a nasty one. The failure mode is:
1. Request A started FFmpeg and waited for a segment.
2. Request B requested an earlier or far away segment.
3. Jellyfin thought FFmpeg should to restart at a different position.
4. Request B killed the existing transcoding job.
5. Killing that job cancelled the same token request A was using.
6. The cancellation produced http 500 to request A.
To fix this:
we lock transcoding job state changes and segment handling per playlist, and use a thread safe counter to track how many http responses are still using each job’s segments. A job is only stopped or replaced once that counter reaches zero.
Prevent FFmpeg's SIDX path from rewriting open-GOP boundary packet timestamps. HLS uses the media playlist for segment indexing and does not require the SIDX box.
* Add support for bitstream filter to remove dynamic hdr metadata
* Add support for ffprobe's only_first_vframe for HDR10+ detection
* Add BitStreamFilterOptionType for metadata removal check
* Map HDR10+ metadata to VideoRangeType.cs
Current implementation uses a hack that abuses the EL flag to avoid database schema changes. Should add proper field once EFCore migration is merged.
* Add more Dolby Vision Range types
Out of spec ones are problematic and should be marked as a dedicated invalid type and handled by the server to not crash the player.
Profile 7 videos should not be treated as normal HDR10 videos at all and should remove the metadata before serving.
* Remove dynamic hdr metadata when necessary
* Allow direct playback of HDR10+ videos on HDR10 clients
* Only use dovi codec tag when dovi metadata is not removed
* Handle DV Profile 7 Videos better
* Fix HDR10+ with new bitmask
* Indicate the presence of HDR10+ in HLS SUPPLEMENTAL-CODECS
* Fix Dovi 8.4 not labeled as HLG in HLS
* Fallback to dovi_rpu bsf for av1 when possible
* Fix dovi_rpu cli for av1
* Use correct EFCore db column for HDR10+
* Undo outdated migration
* Add proper hdr10+ migration
* Remove outdated migration
* Rebase to new db code
* Add migrations for Hdr10PlusPresentFlag
* Directly use bsf enum
* Add xmldocs for SupportsBitStreamFilterWithOption
* Make `VideoRangeType.Unknown` explicitly default on api models.
* Unset default for non-api model class
* Use tuples for bsf dictionary for now
Previously, we predicted whether the subtitle should be burned in with transcode reasons, but that was not accurate because the actual transcoding codec is only determined after the client has requested the stream. This pass through the option to the `DynamicHlsController` to handle the subtitle burn-in during the actual transcoding process. Now the client should be responsible to conditionally load the subtitle when this option is enabled.
Some options that are too performance-intensive are disabled, while
others are left enabled to trade-off between real-time playback and quality.
Signed-off-by: nyanmisaka <nst799610810@gmail.com>
Some Audio codec will have a null CodecTag, check for that to avoid null reference
If the client already requests a specific sample rate, use that instead of our default sample rate for AC-4
Signed-off-by: gnattu <gnattuoc@me.com>
LAME's VBR mode only has advantage over a certain bitrate range. For very low and very high bitrate, use the ABR mode instead.
aac_at's CVBR mode produces very good quality and is not worse than its TVBR mode in blind testing. Use this mode for convenience.
The ffmpeg native aac encoder will have quality regression with its VBR mode. Always use CBR mode for ffmpeg's native aac encoder.
Signed-off-by: gnattu <gnattuoc@me.com>
unkin-agent
Cody Robibero