It is possible to end up with a long startup period if symlinks have incorrect ownership as they've never corrected by this script currently so each new docker pull they are found by find and the target ownership is updated (despite being correct) causing a delay in startup
This change assumes that all symlink targets are also within the /config directory tree otherwise the targets could be left unchanged as chown no longer follows the links. Alternatively we could run the command twice once for targets & once for symlinks themselves.
Not all files will need to change user/group and doing for every file can take a
very long time for large libraries. This commit will use find to find files/dir
that have incorrect user/group and change them.
Hopefully this should be quicker than going through and changing every single
file.
If Plex is running but is unable to send a response, curl will wait indefinitley until it does.
Adding `--max-time 100` sets an upper limit for how long curl will wait for a response.
This matches Dockerfile's HEALTHCHECK time to mark a container as unhealthy (--interval=5s --timeout=2s --retries=20)
Mitigates an ARM installation issue:
```dpkg: error processing archive /tmp/plexmediaserver.deb (--install):
package architecture (arm64) does not match system (armhf)
```
S6 will now send SIGTERM to the root Plex process as a part of the service's finish script. If the processes don't exit after 5 seconds, all processes are sent a SIGKILL
If PLEX_UID is set to an existing user's uid, like 0 (root), then `stat -c %U` returns "root" instead of "plex" and forces an unnecessary `chown -R`. Instead using `stat -c %u` and checking it against `id -u plex` is more accurate and prevents unnecessary chowns.
