Possible to automate preemptible instances? | PlexGuide.com

Possible to automate preemptible instances?

  • Stop using Chrome! Download the Brave Browser via >>> [Brave.com]
    It's a forked version of Chrome with native ad-blockers and Google's spyware stripped out! Download for Mac, Windows, Android, and Linux!
Welcome to the PlexGuide.com
Serving the Community since 2016!
Register Now


Original poster
Feb 25, 2018
In order to get the most value from credits is it possible to automate preemptible instances. With PG version 8.4.x being focussed on GCE I could see two possible stratergies:

1) Have a persistent data store say a 30GB SSD for OS and appdata. Then have 1 / 2 nvme drives in raid which only store downloads as they are uploaded, If none of the persistent data is on the nvme drives such as Sabnzbd history or nzb store, or the sonarr / radarr config and database is safe then upon a reboot the programs should be able to carry on where they left off.

- Possible downside is some paths may need to be changed to better seperate persistent data, Some downloads may end up being downloaded twice depending when the instance is terminated (could be minimized by using sabnzbd duplicate detection), Not really useful as a Plex server.

2) Having the preemptible instance only being a download / upload host with radarr / sonarr being on a permanent (cheaper) host, this might be easier to implement using a ansible playbook at boot time.

I just wanted to throw these ideas out there and get peoples opinions, I feel this would be a great addidtion to the GCE feeder edition.


Oct 28, 2018
With you being able to simply create a new account when credits expire, this really isn't needed. Their really isn't any changes needed to be made as long as the storage is persistent. It will be like shutting down a computer and moving the hard drive around across hardware. As long as there isn't issues with compatibility like drivers, etc it shouldn't be a problem. With GCE you don't need to worry about drivers, etc.

The main concern is the NVME. Because NVME would need to be remounted every time the instance is started up again. Also it being terminated suddenly will really mess with downloads and also be an issue when it comes to backups. There is alot that could go wrong with it, and not much of a benefit to it being added to the main PG branch unless one of us figures out all the logistics and whats needs to be done for Admin to incorporate it.

I am in the same boat for raid automation. I want GCE raid automation in the new version, but setting up different drive configs is a big thing that can break easily. Until I figure out how to script out the different variations and set up some checkers for during installation, it won't be added to PG because there is alot to it for more of a niche use case type setup.

Recommend NewsGroups

      Up To a 58% Discount!