Date: Tue, 19 Nov 2024 00:32:32 -0000 (UTC) From: "Peter 'PMc' Much" <pmc@citylink.dinoex.sub.org> To: freebsd-fs@freebsd.org Subject: Re: Can ZFS ARC be run with roundrobin policy? Message-ID: <slrnvjnn50.2rpc.pmc@disp.intra.daemon.contact> References: <ZpUGt8f6zvZ4Codz@disp.intra.daemon.contact> <fabe073f-44c3-4539-acd6-62f9a7799a31@gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On 2024-11-03, Graham Perrin <grahamperrin@gmail.com> wrote: > On 15/07/2024 12:23, Peter wrote: > >> Hi, >> from what I perceive, the ZFS ARC allocates in firsttouch >> policy, … > >> … not always to my liking. I would prefer roundrobin >> allocations here. But, I didn't easily find a switch, and, reading >> into the UMA allocator, it states that it can do roundrobin, but it is >> not so easily figured how that be made happen in practice. >> >> Pointers on where I might look into, are welcome. > > > Does <https://github.com/openzfs/zfs/pull/14437>; (closed in 2023) help > to shape your thoughts? > >> Specify a balanced NUMA policy for kernel buffers on FreeBSD by >> markjdb · Pull Request #14437 · openzfs/zfs Wow, cool finding! It shows very much the same as what I already did in the meantime, while wondering why it is not even documented in the uma(9) manpage but then seems to work nevertheless.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?slrnvjnn50.2rpc.pmc>