I am seeing about 100% CPU utilization (single threaded, the other 3 vcpu’s are near idle) consistently with PMS running on DS412+ - even when there is not player attached. Looking at the logs there seems to be some issues - things like:
SLOW QUERY: It took 750.000000 ms to retrieve 0 items.
...
Held transaction for too long (../Statistics/StatisticsManager.cpp:245): 0.180000 seconds
Took too long (0.360000 seconds) to start a transaction on ../Library/MetadataCollection.cpp:302
These hit a few times a minute, making me think these are symptoms rather than problems
There is also the occasional issue with SSDP trying to talk to a tivomini device and getting text instead of XML, but again - not often enough to be a root of the issue
Its odd as I do not have a lot of changes in my media - so does it continuously re-scan/re-index same media it already scanned?
As far as CPU, its a dual-core Atom/Celeron - not THAT powerful, but not too wimpy either :-/ But if 100% CPU usage is normal, it is what it is.
As for the issues - it seems that players often are very sluggish, and often disconnects from the server when browsing or searching - although not usually during playback. It also seems to often get confused if I try to skip forwards/backwards. The issues tend to go away after a server restart, only to come back short time later. Things get a lot worse if the NAS is doing something else at the same time.
It may be some IO issue with NAS, but I was hoping to see what PMS is doing to try to understand where to look for bottlenecks. FWIW, same setup with storage being mounted from another NAS seemed to have playback issues, but never browsing/responsiveness issues :-/
For what its worth, rebooting the NAS seems to dropped the cpu usage to around 0-5% - not sure how long that would last, but everything seems happier. Meanwhile I went through and disabled many of the features you listed that I was not using anyway (not even sure what “video preview thumbnails” are for, do not recall seeing that in UI unless that is when it grabs a random image from video instead of the normal cover photo… Thanks
Been over 24 hours, the 100% CPU has not returned so far - FWIW, restarting and even upgrading the PMS app did not help, but restart did… crossing fingers. That said, there is some weirdness going on with this NAS - whenever I do a large network transfer or major IO operation - things get funky.
Other things seems to slow down, during a big file operation all things on the box (not just Plex) get sluggish. If I run PiHole DNS server in a docker container, and it occasionally fails to respond during those times. I discovered the 100% utilization by plex when I started looking at this - but since the reboot, Plex is not at 100% CPU but it is still happening. While in almost all the cases I am transferring over the network, in several cases this happened while unzipping a huge archive locally - so I am not 100% sure if it is IO or network issue.
4x 4TB IronWolfs in a Raid5 (ST4000VN008, about a month old, and yes, they swear up and down that they are CMR - for whatever that is worth) - if I am managing or overload those just by unzipping a file, I’ve done something is very, very wrong(which is not out of the question) :-/
while I know the thread has moved on to a discussion about the HDDs –
If I may offer a few thoughts from personal experience?
The DS412+ uses a D2700 dual core CPU
This CPU has 481 passmarks of total performance capability
It does not meet the 700 Passmark minimum
This NAS can work if used VERY slowly and gently
I used to own one of these CPUs in a DS1813+.
It had SATA-2, not SATA-3. Disk I/O is not quick.
The DS412 is limited at 1GB of RAM. I had 2 GB of RAM and was limited.
In today’s terms, with PMS having grown to the demands it now has, 700 Passmark minimum MUST be taken into consideration.
Were I to make a recommendation:
Choice 1: Use it as the nice little NAS that it is; Nothing more. Put PMS on a host which can handle the load.
Choice 2: Use the DS412 but control how it’s loaded with media, meaning:
Turn off Preview thumbnails during initial media loading (indexing)
Do not add more than 1000 of anything at any one time (movies, TV episodes, Music tracks, or photos). This is done because of the database and lack of PER-THREAD speed with this CPU. (this is the SLOW QUERY: 750 ms root cause.
After adding any block of media – Optimize the database manually.
Repeat the above process until all media indexed (this may take several iterations – expect it and plan for it).
Ugh. Makes sense. Ironically I moved from a (much better PassMark rated) Mac Mini to this NAS and seen a lot of stability improvement, as Mac hardware/software makes for some very frustrating server experience. I guess I need to try it again, or better yet get some other system to be my Plex brain. Still, I do not have a lot of trouble PLAYING media - which is where I expect most of the trouble. I suspect something is wrong with my setup, as when this same DS was running PMS with data being remote mounted from another NAS - it had no problems like this (though occasional playback issues because the other NAS is crappy MyCloud setup)
