Repository navigation
Block pool - #56
Draft
zmc wants to merge 3 commits into
Draft
Block pool#56zmc wants to merge 3 commits into
zmc wants to merge 3 commits into
Conversation
Loop devices can now be backed by slices of a dedicated disk or partition instead of sparse files. - BlockPool enrolls an empty device and hands out fixed-size regions, tracked in a locked state file. It refuses devices that aren't empty or aren't its own. - Released regions are zeroed wherever BlueStore keeps a label copy; otherwise ceph-volume silently skips a reused region. - LoopDeviceMixin picks free loop numbers, safely across parallel creates, and attaches them to pool regions or sparse files. Device access runs on the container host, so this also works through a podman machine. Signed-off-by: Zack Cerza <zack@cerza.org>
TestNode derived loop device numbers from its index, so they could collide with loop devices already in use on the host. Use LoopDeviceMixin instead, which also lets testnodes use the block pool. The stack allocates loop numbers for every container before creating them in parallel, so concurrent creates can't pick the same ones. Signed-off-by: Zack Cerza <zack@cerza.org>
Ship a commented [block_pool] section, add `ceph-devstack block-pool status` to show allocations, and have doctor check that the user can read and write the pool's parent device. Document the pool in the README. On Fedora CoreOS (e.g. a podman machine) the disk group is only defined in /usr/lib/group, where usermod can't add members, so doctor --fix copies it to /etc/group first. Signed-off-by: Zack Cerza <zack@cerza.org>
zmc
added this pull request to stack #57
October 21, 2026 22:18
This branch has not been deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Optionally back loop devices with slices of a dedicated disk or partition instead of sparse files. The pool enrolls an empty device, refuses anything that isn't empty or its own, and clears BlueStore label copies when it takes a slice back. Testnodes switch to the shared loop device code, and
ceph-devstack block-pool statusshows allocations.Testing
On folio12, enrolled a 1.5T NVMe device and ran 12 testnodes from it through
teuthology:no-cephandorch:cephadm:smoke-small. After smoke-small, releasing the slices cleared every OSD's BlueStore labels (checked at +0 and +1G); before that fix, reused slices yielded no OSDs.