Date: Sun, 31 Aug 2025 07:59:55 +0000 From: bugzilla-noreply@freebsd.org To: bugs@FreeBSD.org Subject: [Bug 289204] Numa with 4 domains often fails to boot (zfs related) Message-ID: <bug-289204-227@https.bugs.freebsd.org/bugzilla/>
index | next in thread | raw e-mail
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=289204 Bug ID: 289204 Summary: Numa with 4 domains often fails to boot (zfs related) Product: Base System Version: 14.3-RELEASE Hardware: amd64 OS: Any Status: New Severity: Affects Only Me Priority: --- Component: kern Assignee: bugs@FreeBSD.org Reporter: pmc@citylink.dinoex.sub.org The system gets stuck after starting multiuser boot. The cause in /etc/rc.d/zpool when invoking "zpool import -c <cachefile> -a -N": zpool doesn't return and doesn't compute. procstat -kk shows this stack for that zpool process: mi_switch+0xc3 _sleep+0x205 g_waitidle+0x7c ast_handler+0x8b thread_dtor+0x24 uma_zfree_arg+0x1f0 thread_reap_domain+0x265 thread_count_inc+0x24 thread_alloc+0x16 kthread_add+0xb1 kproc_kthread_add+0xb5 zvol_os_create_minor+0x278 zvol_create_minors_recursive+0xe6 spa_import+0x692 zfs_ioc_pool_import+0xb3 zfsdev_ioctl_common+0x567 zfsdev_ioctl+0x12b devfs_ioctl+0xd1 When manually importing the pools one by one, the problem happens at about every 10th pool (there are more than ten here), apparently randomly. This is 14.3-RELEASE with zpool 2.2.7 Similar never happened with any 13.x release. Also it does not happen with 2 or none numa domains: these boot okay. Machine is ASUS Z10PA-D8 with two E5-2660-v3 chips. The machine runs in Cluster-on-Die mode to separately utilize the two memory controllers of the E5-2660 v3; this gets presented to the OS as separate Numa domains, which makes a total of four. -- You are receiving this mail because: You are the assignee for the bug.home | help
Want to link to this message? Use this
URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?bug-289204-227>