ZFS: Set elevator=noop on disks in the root pool
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
zfs-linux (Ubuntu) |
Won't Fix
|
Medium
|
Colin Ian King | ||
Xenial |
Fix Released
|
Medium
|
Colin Ian King | ||
Zesty |
Won't Fix
|
Undecided
|
Unassigned |
Bug Description
ZFS-on-Linux has its own I/O scheduler, so it sets the "noop" elevator on whole disks used in a pool. https:/
It does not set the scheduler for a disk if a partition is used in a pool out of respect for the possibility that there are non-ZFS partitions on the same disk. https:/
For regular pools, the recommendation is to use whole disks. For a root pools, it's just the opposite. The typical case is that partitions are used. And, for root pools, it is unlikely that the same disks have non-ZFS filesystems.
The debdiff in comment #5 applies cleanly to the latest package and functions correctly. This is an important change for root-on-ZFS users. It has no effect on non-root-on-ZFS installs, because the code is only in the zfs-initramfs package.
Changed in zfs-linux (Ubuntu): | |
importance: | Undecided → Low |
tags: |
added: regression-release zesty removed: patch |
Changed in zfs-linux (Ubuntu): | |
status: | Confirmed → In Progress |
assignee: | nobody → Colin Ian King (colin-king) |
importance: | Low → Medium |
Changed in zfs-linux (Ubuntu Xenial): | |
status: | New → Fix Released |
importance: | Undecided → Medium |
assignee: | nobody → Colin Ian King (colin-king) |
Changed in zfs-linux (Ubuntu Zesty): | |
status: | New → Confirmed |
The attachment "zfs-scheduler. debdiff" seems to be a debdiff. The ubuntu-sponsors team has been subscribed to the bug report so that they can review and hopefully sponsor the debdiff. If the attachment isn't a patch, please remove the "patch" flag from the attachment, remove the "patch" tag, and if you are member of the ~ubuntu-sponsors, unsubscribe the team.
[This is an automated message performed by a Launchpad user owned by ~brian-murray, for any issue please contact him.]