You can cheat about Plex. 
- Create a share and Enter Path Manually. You will find it under
/share/CACHEDEV1_DATA/.qpkg/PlexMediaServer
- From there, you only want a ZIP (or tar.gz if you prefer) of the
Library
directory.
Library contains everything PMS knows about your media.
After you’ve completed your tasks:
- Create the share again
- With PMS stopped
- Remove the exiting contents of Library (you’re going to replace it all anyway)
- Unzip the previous backup into it just like it came out
- Verify you see
.qpkg/PlexMediaServer/Library/Plex Media Server
in the resultant tree structure and you’re good to go
#PlexHatOff
I’m not convinced with all the hoopla about Qtier. Consumer grade SSDs have a very finite life. Put one in a laptop and at the 3 year mark you’re likely replacing it.
Sorry to be an old fart here but after 40 years in the business, slinging around the NAS (even empty), pulling it out of the cabinet where it’s all nicely tucked in, is not something I relish the though of.
Qtier may be nice but when spinning iron, which is actually faster than the singular SATA-3 SSD volume (800+ MB/sec from 7200 RPM vs 520 MB from SSD), and I can change the HD without performing surgery on the unit, the fundamental question of “Why?” hasn’t been satisfied.
Sure, put cache in the 2.5" slots. Got that, sold on that point. Qtier? Nope… not sold yet. The NAS isn’t busy enough to warrant the fuss imho.
- CACHEDEV1 = SSD = QTS core
- CACHEDEV2 = Spinning metal (configured as a thich static volume, inode size set at creation to allow growth to 16 TB HDs without need to recreate volume. (if I get that much stuff, I need a new NAS period)
- 2.5" #1 = 512 GB SSD cache R/W
- CACHEDEV3 = Spinning metal 2.5" 1TB for 'stuff" (which would kill a SSD in a few months)
I don’t really disagree with you. It’s very likely that Qtier is not the best thing since sliced bread, not yet anyway. I also spent 40 years in the business but for me that means I am past the point where I do geeky stuff ‘just for fun’. This server is not a lightweight, as you said. As we get older, slinging around the hardware becomes more of a problem. I want to set up my Plex server with as little future tweaking as possible.
For the moment, I did reconfigure the pool and both tiers are Raid 1, though I ordered a third hard drive so I can migrate to Raid 5.
I did end up having to reinstall Plex because the path you gave me appears to be the backup path…and since my server was less than three days old, the path was empty (no backup had yet been taken). I tried looking for hidden files, but nope…the path was empty.
From now on I can do a decent backup, so I thank you for that info.
Any time i can help, PLEASE call out.
In the ‘Just for fun’ department, I decided to offload all of PMS on to a different SSD (save wear and tear on that M.2). Symlinks to the rescue 
Yea, it’s early days yet for this new QNAP feature but they need to work on it a bit. They should add an option to always force large files (like mp4) to the hard drive to avoid excessive SSD traffic. I really REALLY hated having to reload Plex yet again, knowing that the process was going to beat the hell out of those m.2 drives. But I didn’t hate it enough to make me want to haul the damned thing off the shelf and pull the drives out. I suppose I could have fiddled with the pool settings and exclude them temporarily but at that point I was kinda ‘past it’ if you get my meaning HAHAHA.
That being said, this process might have been a lot easier if Plex itself had an option to force an immediate database backup onto a USB drive. It looks like you can set a time for it but you can’t say ‘please run it NOW’.
hmmmm symlinks…that sounds intriguing. Might have to play around with that a bit. hee hee
They do have the 1MB/ 2MB cache write size limit. I agree it would be nice to make files above size X stay on HD. no need to pull 20GB to SSD
@“mdibella@rochester.rr.com” said:
Sigh. I think I am convinced. No, didn’t have a hard drive failure but if I do, this is a lot of stuff to mess around with. So I suspect I am about to start over. Again. I’ve looked at web pages that describe how to migrate Plex and it says ‘Log on with the console’. Yeah right. Might as well tell me to use my iphone. I haven’t used Telnet in many years…by the time I figure it out again, I could probably just start over from the beginning. I think I may have to delete everything and start from my file backup disk. There is no option to migrate from Raid 0 to Raid 5.
So. My system has two m.2 ssd and two 6 gig HDD. Are you saying I should create two storage pools? One with the SSDs and one with the HDD? And then hand them both over to QTIER? I had everything assigned to one storage pool…again, the documentation on this stuff leaves much to be desired and it’s unclear to me how using multiple pools works with QTIER. I am content to do whatever makes QTIER happy and let it manage this mess.
Oh…wait…If I am reading this correctly, I create 2 tiers and then add both tiers to a single storage pool. ??
(and of course I still have four drive bays empty so I will likely be adding more sooner or later).
For QTier, you basically create 2 Raid-1 Sets (SSD & HDD) and both get assigned to the Storage Pool. QTier then moves blocks within the between the SSD & HDD.
Here is QNAP Video to walk you through it
https://www.youtube.com/watch?v=3bJR8FzRYlw
@ChuckPA said:
They do have the 1MB/ 2MB cache write size limit. I agree it would be nice to make files above size X stay on HD. no need to pull 20GB to SSD
Thing to keep in mind… caching/QTIER is all BLOCK based not file based. The cache threshold is related to transfer block size.
I understand it’s block transfer size. My point is "When playing a 20GB file by 2 different players, will QTIER bring those blocks into SSD because more than 1 device is being streamed to? "
Thanks so much again, guys. I had ordered 2 m.2 drives and 2 6tb hard drives, thinking that would work best…but that was assuming I’d use raid 0. Since raid 5 seems best for the hard drives, I need to buy at least one more. Right now they’re using Raid 1
Then I will have to put my credit card on a diet…
MwC_Trexx: I am about to make you laugh, I bet. Bear in mind I was a software developer for almost 40 years so I tend to think in absolutes. I did find the video you posted and it is well made. I came away with one big question. The whole thrust of the message seems to be ‘assign everything to a storage pool and let QTIER do the heavy lifting’.
Yet in the video, the demonstrator is creating…storage pool 2. Number two. No mention of pool one, though the implication is that only one pool is needed. So the coder who still lives in my head keeps asking, WHAT did he do with pool one? HAHAHA
I am going to add the 2nd M.2 for CACHEDEV1 (QTS & apps) and also put in RAID. They will be some 9 months apart in age which will minimize the chance of both going out at the same time.
Credit card on a diet? Don’t go on a diet, get a bigger belt! 
@MwC_Trexx
I also took an interesting turn in PMS config… It makes maintenance for us ‘seasoned gentlemen’, a lot easier.
- Slide a 2.5" SSD in a slot.
- Relocate PMS’s Library (via symlink) to it.
- Migrate any other comparable apps there as well.
The result is quite profound:
- Much quieter (obviously)
- Much quicker (duh)
- Easier to replace (a requirement)
- No additional wear & tear on the M.2 (the ultimate)
PS: No i will not give up my day job and pursue philosophical writing. You can breathe now 
@ChuckPA said:
I understand it’s block transfer size. My point is "When playing a 20GB file by 2 different players, will QTIER bring those blocks into SSD because more than 1 device is being streamed to? "
My guess would be likely no, because based on my usage at home… real-time isn’t exactly “real-time” (otherwise it would just be CACHE Acceleration) so there does have to be some recurrence of usage for a given data block to make it “hot”.
Now you can pre-dispose shared folders for it to be more aggressive on, but from what I have seen “hot” doesn’t mean it was just accessed.
Now if it was your kids FAVORITE movie and they watched it every day, then it might tend to keep that more “pinned”… but again you also presumably have a bunch of other traffic going in/out of NAS, so HOT is relative.
@“mdibella@rochester.rr.com” said:
MwC_Trexx: I am about to make you laugh, I bet. Bear in mind I was a software developer for almost 40 years so I tend to think in absolutes. I did find the video you posted and it is well made. I came away with one big question. The whole thrust of the message seems to be ‘assign everything to a storage pool and let QTIER do the heavy lifting’.
Yet in the video, the demonstrator is creating…storage pool 2. Number two. No mention of pool one, though the implication is that only one pool is needed. So the coder who still lives in my head keeps asking, WHAT did he do with pool one? HAHAHA
That is a good question 
Based some of his screenshots it looked like he might have had a couple other drives installed and presumable mirrored (SSD/HDD/who knows) which is where DEV1/StoragePool1 went.
Makes sense then.
I guess I’ll stick with my IBM “I’ve allocated 10 cylinders for you. Don’t abuse them” mentality 
@ChuckPA said:
@MwC_Trexx
I also took an interesting turn in PMS config… It makes maintenance for us ‘seasoned gentlemen’, a lot easier.
- Slide a 2.5" SSD in a slot.
- Relocate PMS’s Library (via symlink) to it.
- Migrate any other comparable apps there as well.
The result is quite profound:
- Much quieter (obviously)
- Much quicker (duh)
- Easier to replace (a requirement)
- No additional wear & tear on the M.2 (the ultimate)
PS: No i will not give up my day job and pursue philosophical writing. You can breathe now 
Sounds like a bash script in the making there Chuck 
Do a getcfg for QPKG install location. User passes in new shared folder location as a parameter. Script takes care of the rest 
Personally though even with SSD’s I always go mirrored as I don’t like losing data, although at least in this case it isn’t too painful to backup a shared folder vs. hidden qpkg location.
@ChuckPA said:
Makes sense then.
I guess I’ll stick with my IBM “I’ve allocated 10 cylinders for you. Don’t abuse them” mentality 
Oh I grew up in the semi-old days of NAS, not quite that old though, where you were manually assigning spindles to specific runs, and we didn’t have this new fangled storage pools and auto-striping, and auto-tiering, etc.
I’ll drop a few names. This will show you my heritage.
360/168 Fortran and APL.
DPS-8
Prime 100, 200, 300, PDP-5, Eclipse & Eclipse/MV (production serial number 3), PDP-11/70 and VAX-11/780 (before the 785) + Sun 1