In combination with this suggestion, wouldn't it be better as well to change the default value for Use cache disk to Yes instead of No when creating a new user share AND a cache drive is present?

That goes against what I said in my second paragraph, and seems more complicated to implement. I don't really have any idea what is happening down at the level of shfs so maybe not. A share that is created by specifying /mnt/cache/whatever is still a share but you would want it to be cache-only instead of cache-yes by default. But I guess any that had top level folder(s) on the other disks could be cache-yes by default.

 

I sort of get the idea from reading logs that when the array is started, it reads the config/shares/*.cfg and if any top level folders exist that aren't specified then it "creates" the share in memory, so if a user ever added a cache drive then I guess any share that wasn't explicitly configured would automatically become cache-yes if that were the default.

 

I would definitely settle for leaving the default at cache-no though, and just quit moving folders that aren't explicitly cache-yes.

In combination with this suggestion, wouldn't it be better as well to change the default value for Use cache disk to Yes instead of No when creating a new user share AND a cache drive is present?

 

EDIT:

 

Better?  Sure. 

 

Wanted to update this post because Tom brought up a good point to me when we were chatting on this a few minutes ago.  What if the user only has a single cache device?  To adjust shares to utilize the cache with only a single device present would NOT be a good idea, as it would put data at risk unnecessarily (no redundancy with a single cache device).

 

I also took a good look at all my shares to see how many I use with the cache, and it's actually fairly minimal.  The majority of my shares are set to NOT use the cache because the majority are pulling data over a VPN connection where my bandwidth is constrained enough there that I don't benefit writing to the cache anyhow.  The only shares I use with cache set to yes are my photos and home videos that I copy over the local network.  This isn't downplaying the value of the cache, but rather, highlighting that not all shares warrant the use of cache.

 

The change Tom is proposing is contained entirely to the mover script (no changes to shfs / emhttp) which makes it simple to implement and effective at solving the problem this thread has been discussing.  I think it's solid and support it's inclusion as-is.  Anything beyond that would be another feature-request for future consideration.

How about, instead of this:

 

This change is made in 6.1-rc6, let's see how it works out.