Date: Tue, 5 Sep 2023 01:45:06 -0700 From: Mark Millard <marklmi@yahoo.com> To: Alexander Motin <mav@FreeBSD.org> Cc: dev-commits-src-main@freebsd.org, Current FreeBSD <freebsd-current@freebsd.org> Subject: Re: An attempted test of main's "git: 2ad756a6bbb3" "merge openzfs/zfs@95f71c019" possible file odd result Message-ID: <28614E14-6C55-4183-9315-A098C392B3AC@yahoo.com> In-Reply-To: <6990186B-E85B-4D64-8202-4B9ECBD876FD@yahoo.com> References: <F20BD354-2E10-4CD9-88FB-99DB1E405A52.ref@yahoo.com> <F20BD354-2E10-4CD9-88FB-99DB1E405A52@yahoo.com> <c9c5096b-33d9-1cb6-d446-97b99f324fce@FreeBSD.org> <673A446E-6F94-451E-910F-079F678C5289@yahoo.com> <2BDD30B5-6248-4EC3-83C8-0499E0717D1D@yahoo.com> <C78808EB-FDE2-432D-8309-A7017DFC1BCE@yahoo.com> <69028684-c2cd-8f0c-617f-d0763c08dbe4@FreeBSD.org> <8D832756-6754-4D1E-AE8D-E716FE08F747@yahoo.com> <5605506b-5059-fb72-3e5a-741863d54444@FreeBSD.org> <5724006A-F4C4-42CA-98A9-90C5EB914F5E@yahoo.com> <4E888EC8-4B06-4139-90DF-59AFBA700A5F@yahoo.com> <6990186B-E85B-4D64-8202-4B9ECBD876FD@yahoo.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sep 5, 2023, at 00:00, Mark Millard <marklmi@yahoo.com> wrote: > On Sep 4, 2023, at 22:06, Mark Millard <marklmi@yahoo.com> wrote: >=20 >> . . . >=20 > So I tried 30 for per_txg_dirty_frees_percent for 2 contexts: > autotrim on > vs. > autotrim off >=20 > where there was 100 GiByte+ to delete after a poudriere > bulk run. >=20 > autotrim on: takes a fair time to delete even 1 GiByte of the 100+ = GiByte > vs. > autotrim off: takes less time to delete more. >=20 > The difference is very visible via "gstat -spod" use. >=20 > autotrim on likely makes things less concurrent, somewhat > like USB3 storage having only one command to the device > at a time for FreeBSD. autotrim on seems to prevent the > larger unit of work from being an effective way to > decrease the time required, instead possibly increasing > the time requirement. >=20 > That may be an example of the context dependendency for > what value of per_txg_dirty_frees_percent to use to > avoid wasting much time. Trying autotrim off with 30 for per_txg_dirty_frees_percent got me the oddity/extra-message (just using 32 builders to match the hardware thread count): . . . [00:03:25] [32] [00:00:00] Builder starting [00:03:43] [01] [00:00:18] Finished print/indexinfo | indexinfo-0.3.1: = Success [00:03:43] [01] [00:00:00] Building devel/gettext-runtime | = gettext-runtime-0.22_1 [00:05:20] [01] [00:01:37] Finished devel/gettext-runtime | = gettext-runtime-0.22_1: Success 23/.p/cleaning/rdeps/gettext-runtime-0.22_1/chemtool-1.6.14_4 copy: open = failed: No such file or directory [00:05:23] [01] [00:00:00] Building devel/gmake | gmake-4.3_2 [00:05:55] [02] [00:02:30] Builder started . . . (Not that I know if the context actually matters and I have no clue if I'll ever get a repetition.) =3D=3D=3D Mark Millard marklmi at yahoo.com
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?28614E14-6C55-4183-9315-A098C392B3AC>