rffmpeg
is a remote FFmpeg wrapper used to execute FFmpeg commands on a remote server via SSH. It is most useful in situations involving media servers such as Jellyfin (our reference user), where one might want to perform transcoding actions with FFmpeg on a remote machine or set of machines which can better handle transcoding, take advantage of hardware acceleration, or distribute transcodes across multiple servers for load balancing.
-
Install the required Python 3 dependencies:
click
,yaml
andsubprocess
(sudo apt install python3-click python3-yaml python3-subprocess
in Debian) and optionally installpsycopg2
withsudo apt install python3-psycopg2
for Postgresql support. -
Create the directory
/etc/rffmpeg
. -
Optionally, copy the
rffmpeg.yml.sample
file to/etc/rffmpeg/rffmpeg.yml
and edit it to suit your needs. -
Install
rffmpeg
somewhere useful, for instance at/usr/local/bin/rffmpeg
. -
Create symlinks for the command names
ffmpeg
andffprobe
torffmpeg
, for examplesudo ln -s /usr/local/bin/rffmpeg /usr/local/bin/ffmpeg
andsudo ln -s /usr/local/bin/rffmpeg /usr/local/bin/ffprobe
. -
Initialize the database and add a target host, for example
sudo rffmpeg init && rffmpeg add myhost.domain.tld
. -
Set your media program to use
rffmpeg
via theffmpeg
symlink name created above, instead of any otherffmpeg
binary. -
Profit!
rffmpeg
does require a little bit more configuration to work properly however. For a comprehensive installation tutorial based on a reference setup, please see the SETUP guide.
NOTE Jellyfin 10.10.x and newer require an additional TMPDIR
environment variable set to somewhere exported to the remote machine, or these paths will not work properly. Edit your Jellyfin startup/service configuration to set that. See the setup guide for more details.
rffmpeg
will look at /etc/rffmpeg/rffmpeg.yml
(or a path specified by the RFFMPEG_CONFIG
environment variable) for a configuration file. If it doesn't find one, defaults will be used instead. You can use this file to override many configurable default values to better fit your environment. The defaults should be sensible for anyone using Jellyfin and following the SETUP guide.
The example configuration file at rffmpeg.yml.sample
shows all available options; this file can be copied as-is to the above location and edited to suit your needs; simply uncomment any lines you want to change. Note that if you do specify a file, you must ensure that all top-level categories are present or it will error out.
NOTE: If you are running into problems with rffmpeg
, you must use the config file to adjust logging
-> debug
to true
to obtain more detailed logs before requesting help.
Each option has an explanatory comment above it detailing its purpose.
Since the configuration file is YAML, ensure that you do not use "Tab" characters inside of it, only spaces.
rffmpeg
is a Click-based application; thus, all commands have a -h
or --help
flag to show usage and additional options that may be specified.
After first installing rffmpeg
, you must initialize the database with the rffmpeg init
command.
Note that by default, sudo
/root privilege is required for this command to create the required data paths, but afterwards, rffmpeg
can be run by anyone in the configured group (by default the sudo
group). You can bypass the sudo
requirement with the --no-root
command, for example when running in a rootless container; this will require the running user to have write permissions to the state and database parent directories, and will not perform any permissions modifications on the resulting files.
Once installed and initialized, you can see the status of the rffmpeg
system with the rffmpeg status
command. This will show all configured target hosts, their states, and any active commands being run.
To add a target host, use the rffmpeg add
command. You must add at least one target host for rffmpeg
to be useful. This command takes the optional -w
/--weight
flag to adjust the weight of the target host (see below). A host can also be added more than once for a pseudo-weight, but this is an advanced usage.
To remove a target host, use the rffmpeg remove
command. This command takes either a target host name/IP, which affects all instances of that name, or a specific host ID. Removing an in-use target host will not terminate any running processes, though it may result in undefined behaviour within rffmpeg
. Before removing a host it is best to ensure there is nothing using it.
The rffmpeg
CLI offers a convenient way to view the log file. Use rffmpeg log
to view the entire logfile in the default pager (usually less
), or use rffmpeg log -f
to follow any new log entries after that point (like tail -0 -f
).
If one of the configured target hosts is called localhost
or 127.0.0.1
, rffmpeg
will run the ffmpeg
/ffprobe
commands locally without SSH. This can be useful if the local machine is also a powerful transcoding device, but you still want to offload some transcoding jobs to other machines.
In addition, rffmpeg
will fall back to localhost
automatically, even if it is not explicitly configured, should it be unable to find any working remote hosts. This helps prevent situations where rffmpeg
cannot be run due to none of the remote host(s) being available.
The exact path to the local ffmpeg
and ffprobe
binaries can be overridden in the configuration, should their paths not match those of the remote system(s).
Note that if hardware acceleration is configured in the calling application, the exact same hardware acceleration modes must be available on all configured hosts, and, for fallback to work, the local host as well, or the ffmpeg
commands will fail.
This is an explicit requirement, and there is no easy way around this without rewriting the passed arguments, which is explicitly out-of-scope for rffmpeg
(see the FAQ entry below about mangling arguments).
You should always use a lowest-common-denominator approach when deciding what hardware acceleration option(s) to enable, such that any configured host can run any process, or accept that fallback will not work if all remote hosts are unavailable.
When more than one target host is present, rffmpeg
uses the following rules to select a target host. These rules are evaluated each time a new rffmpeg
alias process is spawned based on the current state (actively running processes, etc.).
-
Any hosts marked
bad
are ignored. -
All remaining hosts are iterated through in an indeterminate order (Python dictionary with root key as the host ID). For each host:
a. If the host is not
localhost
/127.0.0.1
, it is tested to ensure it is reachable (responds toffmpeg -version
over SSH). If it is not reachable, it is markedbad
for the duration of this processes' runtime and skipped.b. If the host is
idle
(has no running processes), it is immediately chosen and the iteration stops.c. If the host is
active
(has at least one running process), it is checked against the host with the current fewest number of processes, adjusted for host weight. If it has the fewest, it takes over this role. -
Once all hosts have been iterated through, at least one host should have been chosen: either the first
idle
host, or the host with the fewest number of active processes.rffmpeg
will then begin running against this host. If no valid target host was found,localhost
is used (see section Localhost and Fallback above).
When adding a host to rffmpeg
, a weight can be specified. Weights are used during the calculation of the fewest number of processes among hosts. The actual number of processes running on the host is floor divided (rounded down to the nearest divisible integer) by the weight to give a "weighted count", which is then used in the determination. This option allows one host to take on more processes than other nodes, as it will be chosen as the "least busy" host more often.
For example, consider two hosts: host1
with weight 1, and host2
with weight 5. host2
would have its actual number of processes floor divided by 5
, and thus any number of processes under 5
would count as 0
, any number of processes between 5
and 10
would count as 1
, and so on, resulting in host2
being chosen over host1
even if it had several processes. Thus, host2
would on average handle 5x more ffmpeg
processes than host1
would.
Host weighting is a fairly blunt instrument, and only becomes important when many simultaneous ffmpeg
processes/transcodes are occurring at once across at least 2 remote hosts, and where the target hosts have significantly different performance profiles. Generally leaving all hosts at weight 1 would be sufficient for most use-cases.
Furthermore, it is possible to add a host of the same name more than once in the rffmpeg add
command. This is functionally equivalent to setting the host with a higher weight, but may have some subtle effects on host selection beyond what weight alone can do; this is probably not worthwhile but is left in for the option.
As mentioned above under Target Host Selection, a host can be marked bad
if it does not respond to an ffmpeg -version
command in at least 1 second if it is due to be checked as a target for a new rffmpeg
alias process. This can happen because a host is offline, unreachable, overloaded, or otherwise unresponsive.
Once a host is marked bad
, it will remain so for as long as the rffmpeg
process that marked it bad
is running. This can last anywhere from a few seconds (library scan processes, image extraction) to several tens of minutes (a long video transcode). During this time, any new rffmpeg
processes that start will see that the host is marked as bad
and thus skip it for target selection. Once the marking rffmpeg
process completes or is terminated, the bad
status of that host will be cleared, allowing the next run to try it again. This strikes a balance between always retrying known-unresponsive hosts over and over (and thus delaying process startup), and ensuring that hosts will eventually be retried.
If for some reason all configured hosts are marked bad
, fallback will be engaged; see the above section Localhost and Fallback for details on what occurs in this situation. An explicit localhost
host entry cannot be marked bad
.
My virtualization setup (multiple 1U nodes with lots of live migration/failover) didn't lend itself well to passing a GPU into my Jellyfin VM, but I wanted to offload transcoding because doing 4K HEVC transcodes with a CPU performs horribly. I happened to have another machine (my "base" remote headless desktop/gaming server) which had a GPU, so I wanted to find a way to offload the transcoding to it. I came up with rffmpeg
as a simple wrapper to the ffmpeg
and ffprobe
calls that Jellyfin (and Emby, and likely other media servers too) makes which would run them on that host instead. After finding it quite useful myself, I released it publicly as GPLv3 software so that others may benefit as well! It has since received a lot of feedback and feature requests from the community, leading to the tool as it exists today.
This depends on what "layer" you're asking at.
- Media Servers: Jellyfin is officially supported; Emby seems to work fine, with caveats (see Issue #10); no others have been tested to my knowledge.
- Operating Systems (source): Debian and its derivatives (Ubuntu, Linux Mint, etc.) should all work perfectly; other Linux operating systems should work fine too as the principles are the same; MacOS should work since it has an SSH client built in; Windows will not work as
rffmpeg
depends on some POSIX assumptions internally. - Operating Systems (target): Any Linux system which
jellyfin-ffmpeg
supports, which is currently just Debian and Ubuntu; Windows might work if you can get an SSH server running on it (see Issue #17). - Install Methods for Jellyfin: Native packages/installers/archives are recommended; a set of Jellyfin Docker containers integrating
rffmpeg
has been created by @Shadowghost. In addition to this special docker image you can use linuxserver's image with this mod. - Install Methods for
rffmpeg
: Direct installation is recommended; a Docker container to act as an ffmpeg transcode target has been created by @aleksasiriski as well as another by @BasixKOR. - OUTDATED Cloud: HCloud Rffmpeg script made to read rffmpeg database and spin up more transcode nodes in Hetzner Cloud.
- Kubernetes: A short guide and example yaml files are available here.
Explicitly no. rffmpeg
is not designed to interact with the arguments that the media server passes to ffmpeg
/ffprobe
at all, nor will it.
This is an explicit design decision due to the massive complexity of FFmpeg. FFmpeg has a very large number of possible arguments, many of which are position-dependent or dependent on other arguments elsewhere in the chain. To implement argument mangling, we would need to be aware of every possible FFmpeg argument, exactly how each argument maps to each other argument, and be able to dynamically parse and update arguments based on this. As should hopefully be quite obvious, this is a massive undertaking and not something that I have any desire to implement or manage in such a (relatively) simple utility.
This has a number of effects:
rffmpeg
cannot adjust anyffmpeg
options based on the host selected.rffmpeg
does not know whether hardware acceleration is turned on or not (see above caveats under Hardware Acceleration), or what type(s) of hardware acceleration are active.rffmpeg
does not know what media file(s) is is handling or where it's outputting files to, and cannot alter these paths.
Thus it is imperative that you set up your entire system correctly for rffmpeg
to work using a "least-common-denominator" approach as required. Please see the SETUP guide for more information.
Explicitly no, though the linuxserver.io docker mod does support this.
I've thought about implementing this more than once (most recently, in response to Issue #21) but ultimately I do not believe this is worth the complexity and delays it would introduce when spawning processes. That issue does provide one example of a workaround wrapper script that could accomplish this, but I do not plan for it to be a part of rffmpeg
itself.
First, run though the setup guide again and make sure that everything is set up correctly.
If the problem persists, please check the closed issues and see if it's been reported before (if it's regarding Emby and you get an "error 127", see Issue #10).
If it hasn't, you can ask in our chat or open a new issue. Ensure you:
-
Enable debug logging in
rffmpeg.yml
(logging
->debug
totrue
) and re-run any failing or incorrect command(s) to obtain debug-level logs for analysis. -
For issues, use a descriptive and useful title that quickly explains the problem.
-
Clearly explain (in the body of the issue or in your chat message) your setup, what is going wrong, and what you expect should be happening. Don't fret if English isn't your first language or anything like that, as long as you are trying to be clear that's what counts!
-
Include your
rffmpeg.log
and Jellyfin/Emby transcode logs as these are absolutely critical in determining what is going on. Use triple-backticks ("```") to enclose logs inline, both in chat and in issues.
I will probably ask clarifying questions as required; please be prepared to run test commands, etc. as requested and paste the output.
Absolutely - I'm happy to take pull requests for just about any bugfix or improvement. There is one exception: please refer to the "Can rffmpeg
mangle/alter FFMPEG arguments?" entry above; unless it's really good work with a very explicitly defined limitation, I probably don't want to go down that route, but I'm more than willing to look at what you've done and consider it on its merits.
I'm always happy to help, though please ensure you try to follow the setup guide first - that's why I wrote it! Support can be found on Matrix or via email at joshua@boniface.me
. Please note though that I may be unresponsive sometimes, though I will get back to you eventually I promise! Please don't open Issues here about setup problems; the Issue tracker is for bugs or feature requests instead.
There's also a fork of this script written in Go with semver tags and binaries available, as well as docker images for both the script and Jellyfin.