Simplest way to setup SSD Cache ......

[TomTiddler]

I have a setup at the moment that has 2 drive pools, call them Pool A and Pool B, each has it's own duplication rules and they work great!

 

I'd like to "front-end" them with a pair of 2TB SSD's , and thought I could create a new pool consisting of the SSD's (Pool C), designate them as SSD's. i.e.non-archive, then add Pool A and Pool B as members of the front-end pool. Will this work?

[TomTiddler]

Should have added the following observation, while setting this up, I removed one drive, and then removed a 2nd drive. Immediately, the first one was shown as "removing", while the second one was marked "Waiting to remove". Frustratingly, I then watched DrivePool happily move files from the first to the second, and then, eventually, moving everything off the second one into the rest of the Pool! Maybe the small tweak to mark the "Waiting to remove" drives as "not available to write on" would be a plan :-)

 

Just saying .......

[Umfriend]

Yes, for multiple drive evac, use of the Drive Usage Limiter balancer is the better way.

I don't think you can add a single Pool as an SSD to two Pools.

Rather, you could partition the two SSDs, say 1TB each partition, pool them together as two 2x1TB SSD pool (so Pool C and Pool D) and then add Pool C as SSD for Pool A etc.

[TomTiddler]

On 11/9/2021 at 3:15 AM, Umfriend said:

Yes, for multiple drive evac, use of the Drive Usage Limiter balancer is the better way.

I don't think you can add a single Pool as an SSD to two Pools.

Rather, you could partition the two SSDs, say 1TB each partition, pool them together as two 2x1TB SSD pool (so Pool C and Pool D) and then add Pool C as SSD for Pool A etc.

not sure what I should be setting in the Drive Usage Limiter - currently, I'm using two SSD's as cache drives for both Pool A and Pool B, and the drive limiter option has both dupllcated and unduplicated boxes checked for both SSD's.

 

[Umfriend]

OK, the Drive Usage Limiter suggestion was in relation to your 2nd post, the on on removing two HDD. Totally unrelated to SSD. On the removal case, there you would uncheck both duplicated and unduplicated and DP will bascally move all files on those disks to disks that may have files.

On SSD, I am not sure I understand. I don't think it'll work the way you suggested it. You can not have a single (set of) SSDs act as SSD for two Pools. In your proposal where you would have Pool C = 2xSSD + Pool A + Pool B, the SSDs would not work as write-cache for any I/O you direct at Pool A or Pool B directly. You *could* of course move all the contents of Pools A and Pool B to the new Pool C-related poolpart.* folders but then all that data is in Pool C and it would be Pool C duplication/placement rules that apply so files may move from Pool A to Pool B and vice versa.

Important to note that when you add Pool A and Pool B to a different Pool C (Hierarchical Pooling), the files in Pool A and Pool B are not in Pool C. Just like when you add a HDD to a Pool, any files that were on that HDD prior to adding, are not in the Pool (until you move/seed them).

Hope this helps.

 

[TomTiddler]

@umfriend - OK, the penny finally dropped, thanks. So, still not sure what my best layout is, but at least I'm pretty sure I understand the behavior better now.

 

Much appreciate the input.