Persistent L2ARC (was: "failed to reclaim memory" with much free physmem)

From: Graham Perrin <grahamperrin_at_gmail.com>
Date: Fri, 12 Sep 2025 20:01:21 UTC 
On 12/09/2025 16:41, Garrett Wollman wrote:
> … Typical machine learning datasets these days are on the order of a
> terabyte, so they won't entirely fit in memory.  (These systems also
> have 2 TB of L2ARC but that gets discarded on reboot, …


Intentionally discarded?

I never worked with a machine learning dataset, but when I did have 
L2ARC: it was always persistent.