Server Version 4.22.3 / Version 1.19.1.2645
Plex Server for whatever reason is constantly writing to the disk (cache or array) and I can’t figure out why. All previous threads on this have been locked as they are over 90 days old.
I was wondering if there was a fix for this?
This appears to be the only file that is CONSTANTLY updated:
/mnt/cache/appdata/PlexMediaServer/Library/Application Support/Plex Media Server/Cache/CloudAccessV2.dat
Hi Ot4mee,
Usually Plex is active in the background as it frequently calls home to know if anyone wants to connect to the server indirectly or for updates. I don’t know exactly why Plex writes to this file, but I can presume it’s not of any performance penalty.
Hey Super4Jet
Performance penalty really isn’t an issue. It’s premature wear of SSD and SMR Drive that I’m concerned about. It looks to be about 300kB/second every couple seconds.
With SMR drive, even a single byte means you have to write up to 300 Megs of data which causes huge delays and performance penalties and excess wear.
With SSD drives, NAND have limited writes, and I would like to try to avoid wearing it out. It also keeps my system drives from spinning down meaning I’m consuming excess electricity.
I wouldn’t be concerned about the SSD, at that write speed it will last centuries, as it is not writing the same block again and again (that’s why built-in capacity > usable capacity). And SMR? I don’t know why you would use an Archive drive as cache.
If you really are concerned, I’ll step back, I never observed Plex in the background.
Even if the SSD last centuries, it still is preventing my unraid from spinning down the disk array which creates excess heat and power usage.
And not everybody has a cache drive. I didn’t until yesterday. Today’s NAS drives are SMR. This is true for just about all the major mfg (WD, Seagate, Toshiba, HGST) unless you go to super large sizes and PRO labels. This creates excess delay, CPU usage, and wear as you have to write up to 300 Megs at a time. So random writes are a disaster performance wise for SMR’s.
I consider it unacceptable. At the very least, use memory to cache the data for a week or a shutdown command, then dump it. It’s a relatively small file. There’s no excuse for it to be constantly updated.
Just saw this and wanted to chime in that for Seagate, none of their Ironwolf (regular or pro) are SMR, as they just had a release about it after the “SMR is bad in a NAS” that somehow everyone just recently figured out; WD Red as you said are SMR, but only the 2/4/6tb sizes, not the 8tb+ ones.
source: 100tb of 4tb Ironwolf, 300tb of 10/12tb Ironwolf, and 120tb-ish of Red 6(boo) and 8’s in service in various phases of projects/upgrades.
Not all WD Red are SMR just the newer drives with EFAX in part number. EFRX aren’t SMR.
This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.
