Created a full Wiki for Installing Unicorn #106
Conversation
added images and wiki.md, edited readme.md
|
Can you review it @drouarb ? |
| #### Assumptions | ||
| 1. You have working linux Plex Server or can set up one quickly with no further instruction. In this guide, we'll be covering PMS deployed in a container. | ||
| 2. You have a working reverse proxy with SSL. In this guide, we'll be using nginx. | ||
| 3. All nodes (UnicornTranscoders) have access to the source media files. This can be accomplish several different ways. Most people will use NFS, SMB, iSCSI, or some other network filesystem protocol tool (such as rclone). |
There was a problem hiding this comment.
It's not mandatory, you can configure your setup so the UnicornLoadBalancer serves the files to the nodes. with CUSTOM_DOWNLOAD_FORWARD. But beware of bandwidth bottleneck
There was a problem hiding this comment.
@drouarb Does that option mean the loadbalencer will serve the raw media files to the node, the node transcodes, and then the node serves the stream?
There was a problem hiding this comment.
According to the docs:
CUSTOM_DOWNLOAD_FORWARD: Enable or disable 302 for download links and direct play, if enabled, transcoders need to have access to media files. Consider changing to true for your needs.
This implies that this option only deals with downloads and direct play (i.e. a direct play where the unicorn software is not normally used will now use the transcoder node to serve the stream). If that's not correct, the docs could use a refresher.
There was a problem hiding this comment.
If my memory is good, this option does that, or maybe there is an other one that does it. @Maxou44 can you confirm ?
There was a problem hiding this comment.
This parameter changes how transcoders access to the media file (for download, direct play, media optimizer and transcoded streams).
If enabled, transcoders access to files on the same mount point, they will read files from theirs own filesystem. FFmpeg will get a local file path.
If disabled (by default) each media access (transcode, optimization...) will call the load balancer to get the file. FFmpeg will get a HTTP link to stream the file using your load balancer.
We recommend to set it to false if your media files are stored on your load balancer server, if you mounted your media files from another service using FUSE or a dedicated storage server, we recommend to set it to true and mount it on each trasncoders (you will save load balancer bandwidth)
There was a problem hiding this comment.
Maybe the docs to should be amended with @Maxou44 's comments. That is much clearer that what the docs state.
| Required Software: | ||
| * NPM/Nodejs | ||
| * git | ||
| You can run Unicorn with Redis, but for a small plex setup with a handful of user/shared accounts, using the built-in sqlite.db will suffice. |
There was a problem hiding this comment.
Redis is for session storage, by default the UnicornLoadBalancer store the sessions in internal memory but this will clear session store if you restart the app.
SQLite is for accessing Plex database.
There was a problem hiding this comment.
Thanks for the clarification.
| 10. Disable outside access to the plex port | ||
| * All requests to the Plex Media Server should pass through the UnicornLoadBalancer. | ||
| * If a user reaches the server directly, the user will not be able to start a stream, since the FFMPEG transcoder built-in to Plex will eventually be replaced by Unicorn's Transcoder. | ||
| * To ensure users always hit the loadbalancer, configure IPTables to drop traffic destined to 32400 from outside your plex machine (drop traffic coming in through the physical nic). You'll need to install iptables-persistent, a package that persists ip tables across reboots, and then set your iptables rule. You must specify your interface, or else 32400 will be dropped completely, even for localhost access, and the loadbalancer will not work. |
There was a problem hiding this comment.
Maybe it would be easier with UFW ?
| ``` | ||
| where `/usr/lib/plexmediaserver` is the binary location. Most mainstream plex containers use this path, such as the container released by LinuxServier.io. | ||
|
|
||
| With that new volume/file mount, restart plex to replace the built-in transcoder with the "fake" transcoder that interprets the transcode arguments. |
There was a problem hiding this comment.
No need to restart, Plex Transcoder is called when you start a stream, if you replace the file, on next stream it would use UnicornFFMPEG instead of Plex Transcoder
There was a problem hiding this comment.
@drouarb We need to restart the container to accept the new volume mount. Otherwise, when plex does call the transcoder, it's still going to have the built in one. Up to this point, the container still has the original. We have not replaced it. So perhaps we can reword this to "restart/run the container with the new volume mount definition"
There was a problem hiding this comment.
Oh yes, I didn't understood that you were talking about a docker setup
| 2. `git clone https://github.com/UnicornTranscoder/UnicornTranscoder.git` | ||
| 3. `cd UnicornTranscoder` | ||
| 4. `npm install` | ||
| 5. `npm run install` |
There was a problem hiding this comment.
This step is only mandatory as root because npm has a security feature that prevent the install script to run on npm i when run as root.
| 5. `npm run install` | ||
|
|
||
| The transcoder is now ready to run. Just like the loadbalancer, we're going to run this as a service. Be sure to substitute the envrionment variables for your own setup/FQDN: | ||
| 1. `nano /etc/systemd/system/unicorntranscoder.service` |
There was a problem hiding this comment.
Maybe names of services should be consistent earlier UnicornLoadBalancer was camelcase
There was a problem hiding this comment.
This is the result of a "Find and Replace" to fix a spelling issue. Easy enough fix
| User=root | ||
| Group=root | ||
|
|
||
| Environment=TRANSCODER_DEBUG=true |
There was a problem hiding this comment.
This shouldn't be enabled by default, it produces a lot of log
| If you want to adjust the envrionment variables in `config.js`, make sure you re-run the npm build with `npm install` & `npm run install`. If you opt to override these variables in the service (like we did above), edit the service file. If you edit a service after it has already been loaded, you need to reload the daemon with `systemctl daemon-reload` & `systemctl restart unicorntranscoder.service`. As a rule of thumb, if you use service environment variables, don't edit the config.js variables (or vice versa). Keep all your variable definitions in the same place. | ||
|
|
||
| Once the FQDN is up and active with working SSL, point your browser to it. It will look something like this: | ||
|  |
There was a problem hiding this comment.
Maybe we could store the image somewhere else than the code repository ?
In this pull request, I created a wiki. Currently, the instruction set for the "UnicornSuite" is dispersed throughout the three repositories. This wiki brings them all under 1 cohesive guide for new/intermediate users. I also covered a few "gotcha" details that are discussed in a few of the closed issues, but not in the readme.
Changes:
Added wiki.md
Edited README.md to link to the wiki
Added an images folder with a screenshot