I just migrated from a 2TB spinning disk to a pair of 500GB Samsung 750 EVO SSD'sfor my cache pool. I have my docker image, 1 vm, appdata and my downloads go there. Ive noticed now that when its under what i assume is heavy load (lots of downloading, extracting, etc) that everything becomes sluggish. Navigating through the unraid gui, playing videos using plex or emby and just everything in general. Once what i assume to be heavy load is done everything returns to normal. With that said that brings me to a couple questions:

 

1. Is there a way to increase performance here?

2. Is there a way for me to add another drive outside of the array/cache pool where i could just have downloads, etc go so the cache pool isnt garbage?

3. Should i not have all of the items listed above on my cache pool for some reason?

1. Unsure without more info, logs etc. Have you left a core free for UnRaid to use? (Preferably core 0)

2. You could use the Unassigned Devices plugin (Available from CA, see below) to do this but there is no issue with what you are doing currently.

3. Nope, what you are doing is fine.

 

First off, if you haven't already, install the Community Applications plugin and then from there Install the Fix Common Problems plugin. This should highlight any obvious issues in your config.

The 750 Evo is a TLC. Presumably you have them for a while so they are saturated.

Once a TLC SSD is saturated, things will be slowed down drastically.

 

1. Install the Trim plugin (just search it in CA) and set it to run periodically (I set it to run daily - see the Scheduler).

2. Unassigned Device. Especially considering it's temp content, probably that's best as you don't really need RAID1 protection presumably.

3. I have everything on cache with no problem. But I use 850 Evo which is 3D NAND so it doesn't crash as drastically as traditional TLC.

@NeoDude - Diagnostics are attached, where would i check the cores?

 

@testdasi - the drives are brand new, how often would i want to run trim then?

storage-diagnostics-20160807-0935.zip

I'm assuming you have dockers and VMs running. When you assign cores to these it's a good idea to leave the first couple of cores free for UnRaid to use. Checking through your logs just now...

Ok, so I noticed there was some issues your docker share? You had it set to Cache only but there was also files on the array in this share. Have these issues been fixed?

I was not aware of any issues but will take a look.  Also,  when you were mentioning cores you mentioned docker,  do I need to exclude specific cores for docker add well? If so,  where do I do that?

The issues were flagged in the "Fix Common Problems" plugin. You might want to look there first.

 

For the cores issue have a read through of this thread regarding Docker Core Pinning.

Ah,  ok.  I ran the plugin then got pulled away and just now saw it.  Will take a look at it. Thanks for the info sir.

No problemo

@testdasi - the drives are brand new, how often would i want to run trim then?

How brand new is brand new?

Your wear levelling counts are 84-89 for the 2 ssd so they definitely were under heavy use for at least a while

IIRC, 84 wear levelling count means about 40TB of data has been written to the disk.

I schedule trim run daily for my ssd.

New as in they showed up last week and i installed them.

New as in they showed up last week and i installed them.

There are 3 possibilities I can think of

[*]These drives were put under heavy activity. That's quite possible if you run download + extracting + copying 24/7.

[*]You bought used and/or got shafted by the seller

[*]My understanding of wear levelling count for Samsung SSD is wrong.

Or a combination.

 

In any case, I recommend running a trim.