I am thinking of using 2 x 2TB HDDs and 1 x 120GB SSD in the default RAID1 configuration in a cache pool. The aim is three-fold:

1. to avoid Plex spinning up the array when not actively being used;

2. to provide a large download area for Transmission off the array; and

3. to provide some redundancy against failure of a disk.

 

Does my proposed cache setup meet those goals? I guess my big assumption and hope is that disks in a Btrfs RAID1 cache pool are able to spin up and down independently, but this may not be possible because of striping? If my assumption is wrong, then would forgoing the SSD (which I have to buy) and just living with constantly spun-up 2 x 2TB HDDs (which I already have on hand) be the way to go?

Edited April 19, 20188 yr by kimguan

19 minutes ago, kimguan said:

not be possible because of striping

this^

 

Another approach would be to reconsider how you would use cache. You could just use a single 120GB SSD as cache for dockers for example, and not cache any user shares. CA Backup could backup your appdata to the array nightly and redundancy wouldn't be important.

 

Many people just starting out jump on the idea of cached writes to user shares when that might not really be important depending on the way they use their server.

 

Most of the writes to my server are queued downloads, scheduled backups, and other unattended processes, so a little slower write speed isn't important to me since I am not sitting there waiting on it to finish anyway. And since these get written directly to the array they are protected by parity, and they don't have to get moved to the array later.

 

 

 

Thanks trul! Looks like I'll have to reconsider my "solution" to this...

Edited April 19, 20188 yr by kimguan