Re: zpool/zfs-commands hang; ZFS filesystem/ZVOLs still running.
- In reply to: Eric Borisch : "zpool/zfs-commands hang; ZFS filesystem/ZVOLs still running."
- Go to: [ bottom of page ] [ top of archives ] [ this month ]
Date: Fri, 17 Jun 2022 18:33:52 UTC
On Fri, Jun 17, 2022 at 10:46 AM Eric Borisch <eborisch@gmail.com> wrote:
>
> Good morning,
>
> I have a 13.1 system where twice now (required hard restart last time
> to resolve) since upgrading from 12.x I have had zfs/zpool-command
> operations start to hang indefinitely, while the underlying pools are
> still actively running/processing data via both filesystem and device
> interfaces.
I went back to some data I captured during the first instance, and it
also had a zfs destroy and zfs send in perpetual D-state:
11076 - D 0:00.01 zfs destroy -d
kspace/users/eborisch@zfs-auto-snap_frequent-2022-06-13-08h15
16564 - D 0:22.82 /sbin/zfs send -L -RI @MARK-mr-dewey
kspace@NEW-mr-dewey
11076 190929 zfs -
mi_switch
_sleep
rms_wlock
zfs_umount
dounmount
zfsctl_snapshot_unmount
zfs_ioc_destroy_snaps
zfsdev_ioctl_common
zfsdev_ioctl
devfs_ioctl
vn_ioctl
devfs_ioctl_f
kern_ioctl
sys_ioctl
amd64_syscall
fast_syscall_common
16564 190128 zfs -
mi_switch
_sleep
rms_wlock
zfs_umount
dounmount
zfsctl_snapshot_unmount
dsl_dataset_user_release_impl
dsl_dataset_user_release_onexit
zfs_onexit_destroy
zfsdev_close
devfs_fpdrop
devfs_close_f
_fdrop
closef
closefp
amd64_syscall
fast_syscall_common
Note the particular @MARK- and @NEW- snapshots are different now, as
they are constantly destroyed/renamed during successful backup cycles,
which have occurred between these two events.
Thanks,
- Eric