https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=277389
--- Comment #44 from Mark Millard <[email protected]> --- Overall: ) Not replicated ) But also not a matching test context Some details: I no longer have the same context available to test. For one, the ZFS media is smaller and the original test files would leave almost no space on the media, given what else would be there. So scaled to be more similar to my fraction from before. (Previous tests were not of nearly-full zfs pool. All tests, old and new, are of a pool in a single partition on a single media.) I tried a little trading number of files/processes vs. file size. I can control the RAM via /boot/loader.conf . I tried 16 GiBytes of RAM and 23 GiByes of RAM. (NO SWAP.) A read test takes under 1.5min for a non-debug kernel for the read test min for the M.2 interface to a U.2 Optane and the additional cores (that are older/slower). That is much shorter than the original report but longer than some of the others. A read test takes under 3min for a debug kernel. It is an officially distributed pkgbase main boot kernel and boot world, no chroot or jail involved. So far, the non-debug kernel based tests have not failed. Nor have the debug kernel boot based tests failed. # uname -apKU FreeBSD CA72-16Gp-ZFS 16.0-CURRENT FreeBSD 16.0-CURRENT main-n285786-0f6c880fcecd GENERIC arm64 aarch64 1600018 1600018 Most recent snapshot info: INSTALLED: Via pkg-static info -C -x ^FreeBSD-kernel- . . . 1 FreeBSD-kernel-generic-nodebug-16.snap20260510045638 1 FreeBSD-kernel-generic-mmccam-16.snap20260510045638 1 FreeBSD-kernel-generic-16.snap20260510045638 INSTALLED: Via pkg-static info -C -x '^FreeBSD-' ( but no FreeBSD-set- ) . . . 8 FreeBSD-*-16.snap20260510045638 . . . -- You are receiving this mail because: You are the assignee for the bug.
