Re: git: 730ea72c706e - main - cam(4): Limit search for disks in SES enclosure by single bus
Date: Tue, 05 Oct 2021 19:54:58 UTC
On 05.10.2021 15:48, Alan Somers wrote: > On Tue, Oct 5, 2021 at 1:41 PM Alexander Motin <mav@freebsd.org> wrote: >> >> On 05.10.2021 15:14, Alan Somers wrote: >>> On Tue, Oct 5, 2021 at 1:04 PM Alexander Motin <mav@freebsd.org> wrote: >>>> >>>> The branch main has been updated by mav: >>>> >>>> URL: https://cgit.FreeBSD.org/src/commit/?id=730ea72c706ef8e025616772cfd86fd89ed3d42e >>>> >>>> commit 730ea72c706ef8e025616772cfd86fd89ed3d42e >>>> Author: Alexander Motin <mav@FreeBSD.org> >>>> AuthorDate: 2021-10-05 19:01:16 +0000 >>>> Commit: Alexander Motin <mav@FreeBSD.org> >>>> CommitDate: 2021-10-05 19:01:16 +0000 >>>> >>>> cam(4): Limit search for disks in SES enclosure by single bus >>>> >>>> At least for SAS that we only support now disks are typically >>>> connected to the same bus as the enclosure. Limiting the search >>>> scope makes it much faster on systems with multiple buses and >>>> thousands of disks. >>>> >>>> Reviewed by: imp >>>> MFC after: 2 weeks >>>> Sponsored by: iXsystems, Inc. >>>> Differential Revision: https://reviews.freebsd.org/D32305 >>>> --- >>>> sys/cam/scsi/scsi_enc_ses.c | 8 ++++++++ >>>> 1 file changed, 8 insertions(+) >>>> >>>> diff --git a/sys/cam/scsi/scsi_enc_ses.c b/sys/cam/scsi/scsi_enc_ses.c >>> >>> >>> Sounds juicy. Do you have any benchmarks? >> >> I don't think we've benchmarked that system with payload yet, if that is >> what you mean, polishing usability. But I don't think there can be too >> much interesting, since big speeds don't require so many HDDs these >> days, but much smaller number of SSDs instead. This particular system >> is about capacity -- ~20PiB raw on 1248 disks in one ZFS pool. :) > > 1248 disks? That's huge. But I'm not asking about throughput. Your > commit message said "much faster". What operation is "much faster"? > Enumerating disks at startup? That would be useful. I don't think you notice the difference on startup, too many other things going on, but during idle it reduces periodic CPU load spikes by enc_daemon processes by several times, proportionally to the number of HBAs/buses. More annoyance than a real problem though. -- Alexander Motin