Server Version#: 1.14.1.5488
Player Version#: 3.77.4
After starting to migrate Plex over to a different server install (I had no idea what I was doing so I used default partition layouts for fedora server, dumb mistake. Also 20tb of data had to be moved at 1Gbps ) I started to move all database files for the entire library along with account data, user accounts, play history, etc. And I noticed it was taking along time to move the Plex DB files, so I went to check how much has been moved and we’ll the default allocation size is definitely not efficient for the Plex DB.
Size: Actual size of the data
Size on disk: Size of space used in disk
It sounds like you have Chapter and Thumbnail images turned on. They will consume a lot of data.
You’re also looking at it through a GUI Windows is showing you the sum of the file sizes versus and the sum of the allocated size. On Linux, as well as Windows, the smallest unit of allocation is 4K bytes.
I use Fedora as my daily driver and configure 128GB for root, default boot, 32GB swap, and the rest for home.
I recommend we put your /var/lib/plexmediaserver in /home/plexdata where you hopefully have the largest space allocated.
Configuring Plex to look for the data there instead of */var/lib/plexmediaserver is a simple step.
How do you have your data allocated currently?
Is it in a free standing partition / device?
Yeah I do know I was looking at it through windows as I have the Plex user data shared locally on the network for ease of access. Its also all in a temp folder so i can move it all at once off the server to my windows desktop
My partition scheme looks like this:
Device Start End Sectors Size Type
/dev/sda1 2048 4095 2048 1M BIOS boot
/dev/sda2 4096 2101247 2097152 1G Linux filesystem
/dev/sda3 2101248 64465137663 64463036416 30T Linux LVM
Where all the data is on the 30TB array
My plan is to partiton it like this:
Device Start End Sectors Size Type
/dev/sda1 2048 4095 2048 1M BIOS boot
/dev/sda2 4096 2101247 2097152 1G Linux filesystem
/dev/sda3 2101248 2149584896 2147483648 1T Linux LVM
/dev/sda4 ? ? ? 29T EXT4
so that in the event that I do want to install the OS agian (im trying alot of exparimentl automation with it), I dont need to move terabytes of data over a 1Gb network again. im going to try cuting the allocation unit size in half and see how that works as the allocated size is almost twice that of the actuall size. Might go on its own partiton so that i can have a larger unit size for the actuall video files.
Partiton size was never a problem for me, its a fedora server with a 12x3TB raid 5+0 (I think, i know its 0+5 or 5+0, I can loose 2 disks and everything is still okay), its the actuall allocation unit size. thanks for you insight tho
Hopefully it’s a RAID 5 at minimum and not a RAID 0 (concatenated only) volume.
What strike me as odd, is how it’s assembled. It is my experience, mdadm presents the new LV as a new device ID (e.g /dev/sdb) . Perhaps it’s just my way of doing things. I had always installed the OS such that it’s standalone and disconnected from the data. That gave me the freedom to reinstall/repartition it if ever needed.
You will be however, VERY hard pressed to put everything you need in 1GB seeing that the default *boot partition size allocation is 1GB. I can’t see any way around that. Even those with 16 GB partitions (what fedora used to do) is too little. PMS itself expands to 250MB. For you to have any appreciable server software and/or desktop GUI (Gnome) just won’t fit.
Here is my boot SSD. It is configured for dual boot (install Windows first trick). I gave each OS 500GB, which is more than enough.
chuck@lizum ~.189]$ sudo parted
GNU Parted 3.2
Using /dev/sda
Welcome to GNU Parted! Type 'help' to view a list of commands.
(parted) p
Model: ATA Samsung SSD 850 (scsi)
Disk /dev/sda: 1000GB
Sector size (logical/physical): 512B/512B
Partition Table: msdos
Disk Flags:
Number Start End Size Type File system Flags
1 1049kB 525MB 524MB primary ntfs boot
2 525MB 501GB 500GB primary ntfs
3 501GB 502GB 551MB primary xfs
4 502GB 1000GB 499GB extended
5 502GB 639GB 137GB logical xfs
6 639GB 656GB 17.2GB logical linux-swap(v1)
7 656GB 1000GB 344GB logical xfs
(parted)
Servers are a pain to do anything but network boot XD anyhow, finally got my partitions set up, before I go ahead just wanted to see if you had any recommendations. Screenshot attached
I’ve removed some of the local user tmpfs filesystems.
This is a CentOS7 host, dedicated to Plex with Sonarr, Radarr, NZBGet and tvHeadend/tvhProxy, small SSD running the operating system (centos_plex VG,) and a 1tb rotating drive for downloads. I’ve also moved the Plex transcoding temp directory under /opt/nzbget/downloads (yeah, I know… but it was expeditious at the time. I may change it.)
Speaking personally, as a *ix nerd for almost 3 decades, I really dislike the Linux tendency to go towards fewer large filesystems. /var should be a separate filesystem, used for variable runtime stuff. Likewise, /opt & /usr should be broken out. Applications should not be in a position to fill up a filesystem and bring the host down. Granted, it’s much less a problem now than in years past, when applications ran in an effectively collapsed server farm on a single node (database, application, etc. all running on 1 machine.)
On a side note… I’d love to run Plex on openSuSE, but everything that I’ve read indicates that there’s some packaging issues (I am NOT pointing a finger here!) that leads to some rather significant pain. I do thoroughly dislike the Fedora/CentOS/RHEL/OracleLinux installers.
Yes, your background shows.
No, you do not need anywhere near that level of subslicing in this arena.
Domestic & private server configurations run surprisingly problem-free with the 4 basic paritions.
Boot
Root
Swap
Home
If you want to make home smaller and create a 5th for opt go for it but very little installs out there. Anything landing there will be because of a customized configuration.
Also, I’ve found that EXT4 covers most cases but if I have any concerns, XFS removes all questions. Personally, I would love to have XFS on the NAS. There would never be any concerns about filesystems > 100 TB. (EXT4 is ok at good for 50, fair-to-OK at 100 but starts becoming unsafe much above 100T)
From one tri-decade nerd to another? Lighten up on the reigns. Loosey-goosey here really does work super in these cases plus, I’m just too darn cranky and tired of constantly redoing filesystems / configurations. I want it done once and only once. We’ve worked harder, now it’s time to work smarter.
As for the openSuSE packaging, I fixed it
Several months ago to be precise. One of our guys found a major issue. I was asked to investigate. I took a look was exasperated at what I found. I rewrote it from ground up. I am finding out now where that went. Thank you for letting me know. I will post back on that.
My rewrite of it is in place to deploy as all the new packages deploy in these next couple weeks. (we did a trial deployment today to work out any last minute kinks and found a couple so that will be done in another day or two)
As I rewrote it, I drew the same line which I did with all the other Linux distributions. OpenSuSE 15 and above are supported because they use systemd as the init program. To check yours, simply type cat /proc/1/comm. It will return init or systemd. If it returns systemd you are good to go. If not, I urge you to update to current.
No argument that home & small systems like Plex don’t necessarily require an enterprise build-out… I also wish I had a dollar for every time a host in the office has crashed and gone down hard because of stupid user tricks (engineered systems, dontcha know… they know better than I…)
If you’ve got the openSuSE issues worked out, let me now and I’ll blow it into a VM. I’ll beat it into the ground.
I’m also seeing tonight that I may have been throwing rocks at the Plex installer (final comment.) I really dislike the Fedora/CentOS/RHEL/OL operating system installers. I’ve gotten spoiled with openSuSE’s installer.
It will be the same metadata configuration as the other Linux distros:
/var/lib/plexmediaserver (which I really wish I could change)
With systemd, you can easily put the data wherever you like.
You only need create the service override file (override.conf) and specify the new environment variable… Done.
) I started to move all database files for the entire library along with account data, user accounts, play history, etc. And I noticed it was taking along time to move the Plex DB files, so I went to check how much has been moved and we’ll the default allocation size is definitely not efficient for the Plex DB.
