I recently re-enabled my cache drive after running my Unraid box without one for some time. I moved everything to the cache drive (docker, appdata, shares, etc.) and the mover runs once in the morning. I also added a 2 hour spin-down delay.

 

What I can't seem to figure out, is why I have a consistent write happening to drive 3 (see below) that paces just under 100 writes per minute. If I clear the stats on the Main page, drive 3 consistently gets a few writes added to it a second. This happens all day and night, regardless. Needless to say, that drive never spins down. 

 

 

Here is a snapshot of the active dockers:

 

 

- All shares are set to use the cache

- All docker settings run through /user/ and not the direct disk

 

How can I figure out what is slowly writing to the third drive?

You could go old school and simply turn off all your Dockers and see if there are any writes. If there isn't turn one on at a time and watch until you identify... 

 

or 

 

You could install either the "Open Files" or "File Activity" Plugin. I personally use the "Open File" one, but I'm sure they would help you out.

11 minutes ago, kizer said:

You could go old school and simply turn off all your Dockers and see if there are any writes. If there isn't turn one on at a time and watch until you identify..

 

Ha! I love how we often get so focused on the technical solution, that we can overlook the basics. Testing... will return with results.

Edited October 25, 20178 yr by AnnHashaway
Grammar

OK, so I had some tidying up to do in order to put the main work on the cache drive.

 

1. I had to manually move the appdata folder from each disk, directly to /mnt/cache/appdata and set the appdata share to Cache:Only. Used command line: 

cp -r /mnt/diskX/appdata/* /mnt/cache/appdata/

 

2. Turn off each docker and turn them on one-by-one, while resetting the Main page stats so I could see which dockers were accessing the data drives consistently.

 

This resulted in zero writes happening to the data drives, but I was still getting a few errant reads that kept a few of the drives up. It turns out that the apps Deluge and Sonarr were the culprits. Deluge is going to access a data drive any time it is uploading to someone else. Sonarr is a bit trickier, but basically, it is constantly checking the selected shared folder where your download programs would drop a file that was ready to be moved/renamed.

 

I have an additional SSD plugged in and sitting idle. Its mounted via the Unassigned Devices plugin, and I think I can basically make part of it a seedbox, so Deluge will only pull data from that and keep the data drives from spinning up.

 

I haven't worked through a solution for Sonarr yet, and it seems to access the shared folder every minute or two, so it keeps whichever drive Unraid is using for it up all the time. I think I may be able to create a new share that only lives on the cache, and make that the shared folder for Sonarr and the download programs. Any other ideas for this?

 

I'll report back with my results.

 

Edited October 27, 20178 yr by AnnHashaway

For future reference, instead of setting those shares to cache-only and then moving the files manually, you could have set them to cache-prefer, and mover would have moved them for you.

 

unRAID may have to spin up a drive when writing just to check if the file already exists on one of the drives.