So one of my ventures is hosting internet radio servers for people to broadcast with. Some people may be familiar with a piece of software called CentovaCast, but they’re way too non-transparent for my liking, plus they haven’t updated in forever, so I choose to use a FOSS software called Azuracast for my accounts - it handles the encoders, playout, automation, playlists, media, the whole bit…
I had moved the installation through about 5 different servers, the latest being a 4.5TB storage server on Servarica that basically increases by 6GB a day, and takes effect when you reboot the server. We started having issues where when you looked for an artist on the back end database, say you searched for Fleetwood Mac… it would not display any results and just kind of error out. I diagnosed the problem down to bunch of partial uploads in the database that were just kind of “hanging there” with no physical files present. I ended up not being able to fix it via hacking with software like dbeaver, so I opted to recreate the playlists and media folders.
One of the other things I wanted to do besides fix that error was switch to a version of Azuracast that doesn’t run in a docker container. The supported install is docker, but there is an ansible installation available and I wanted to use that. So I was having issues with the autodj service not starting for whatever reason. Log mining pointed to a master_me compression setting used to normalize volume of tracks between songs. Shut that off and everything was groovy. Going to move it over to production in the next couple of days.
Just wanted to post this to show that persistence and patience pays off with Linux based toys sometimes… 