Scale 50% use of ARC

[NugentS]

I understand that this is an upstream issue and that scale is a work in progress but there is one thing that bothers me, perhaps even irritates me (slightly).
That is the Linux ZFS only uses 50% of memory (its not actually that bad when there are loads of additional services.

256GB of RAM and I get:

I am running a VM with 16GB running portainer
I am running 4 apps on K3S (Storj, Filebrowser, Netdata & scrutiny)
I am running a Scale-Jail with debian, docker and portainer. (https://gist.github.com/Jip-Hop/4704ba4aa87c99f342b2846ed7885a5d). 6 containers, but one of them is a memory hog that I have "limited" to 40GB

But I (TrueNAS) am still leaving 40GB hanging.

So my question - to IX really (but anyone else that may know something) is there work going on in the background to improve this or is this something we will just have to get used to in the long run.?

And no I cannot (don't want to) use zfs_arc_max to set a different value than 50%. Its not honoured anyway (or rather it gets reset quickly back to 50%, I suspect when VM's etc start up). Also given that this is my server - just for me - I suspect that the extra 40GB ARC will make little difference

 

[morganL]

I understand that this is an upstream issue and that scale is a work in progress but there is one thing that bothers me, perhaps even irritates me (slightly).
That is the Linux ZFS only uses 50% of memory (its not actually that bad when there are loads of additional services.

256GB of RAM and I get:
View attachment 66734

I am running a VM with 16GB running portainer
I am running 4 apps on K3S (Storj, Filebrowser, Netdata & scrutiny)
I am running a Scale-Jail with debian, docker and portainer. (https://gist.github.com/Jip-Hop/4704ba4aa87c99f342b2846ed7885a5d). 6 containers, but one of them is a memory hog that I have "limited" to 40GB

But I (TrueNAS) am still leaving 40GB hanging.

So my question - to IX really (but anyone else that may know something) is there work going on in the background to improve this or is this something we will just have to get used to in the long run.?

And no I cannot (don't want to) use zfs_arc_max to set a different value than 50%. Its not honoured anyway (or rather it gets reset quickly back to 50%, I suspect when VM's etc start up). Also given that this is my server - just for me - I suspect that the extra 40GB ARC will make little difference

For Linux, its the safe margin for ARC due to fragmentation on SLUB allocator.
We're treating it as a bug and plan to fix or mitigate it....
Planning to fix in Cobia.

 

[NugentS]

@morganL - that is good to know - thank you