Have been running truenas scale for some months now, always no issue but now I was getting performance issues when accessing data, so had a look at the memory dashboard and to my surprise, services was eating up 50 Gb of ram.
So I’m running:
One VM running nixos, assigned statically 16gb of ram (in the screenshot above, I had just turned that VM off to see if that was the issue)
SMB & NFS, usually with no load except in the evenings when I’m streaming video - I turned that off as well to see if it would bring memory usage down, but no effect
1 application: tailscale. Turned that off and also unset the application pool to turn off application service but no change.
Went into htop, I see a lot of middlewared processes
Tried service middlewared restart, even just stopped it, but no change in memory.
Anything else I can try? After a reboot, the services ram drops to normal usage again, but it creeps up over a few days until it reaches around 58-59 gbs, only leaving 1-2 gb of ram for my ARC cache…
@kris Here is a new screenshot of htop with userland processes hidden. I had to restart truenas since it force quit my VM because it was running out of memory… In this screenshot, truenas has been up for about 2 days, services ram has expanded in that time from 17gb to 25gb. Running 1 vm with 16 gb, 1 app (tailscale) and nfs/smb/ssh services.
I might be kicking a dead horse here, but I’m having the same issue. In my case, asyncio_loop is using about 24-30GB of my ram. Did you manage to fix it?
I’m using 24.04.2. I even submitted a bug report over on Jira. Only joined in on the topic, since it’s the same process that’s killing my rig. Was wondering if a solution was found.
Otherwise try updating?