Releases: cloudfoundry/grootfs
Releases · cloudfoundry/grootfs
0.17.1
...this patch release:
- Improves the performance of the graph cleanup store measuring operation.
- Sets the BOSH release Dropsonde port to 3457.
- Changes
grootfs delete
in order to not corrupt the store when it deletes a half-created image.
0.17.0
- Enhanced metrics: https://github.com/cloudfoundry/grootfs#metrics.
- New BOSH release property:
grootfs.skip_mount
. grootfs create
JSON output:- Added a boolean BOSH release property named
grootfs.json_output
to makegrootfs create
calls return JSON. - This property will be removed in the next releases.
grootfs create
will only return JSON output.
- Added a boolean BOSH release property named
v0.16.0
Updating GrootFS to v0.16.0, if running with GrootFS already, will require recreating the Diego cells.
grootfs create
can now, optionally, return a JSON object instead of a rootfs path. Use--json
orcreate.json
to enable this feature. This feature flag will soon be removed and JSON output will be the default behaviour.grootfs create
can now, optionally, not mount the final container filesystem. Use--without-mount
orcreate.without_mount
to enable this feature.- Fix: Pull images with opaque whiteouts while using user-mappings and run by root.
Overlay-XFS filesystem driver
- Support images with more than 40 layers.
- Rootless create of local tar-ball-based images without disk quotas.
BTRFS filesystem driver
- Pull images with opaque whiteouts while using user-mappings and run by a non-root user.
- Delete BTRFS subvolumes created in the container before deleting the image.
init-store/delete-store work
grootfs init-store
can create a store based on an existing and mounted BTRFS/XFS filesystem.grootfs init-store
receives --uid-mapping/--gid-mapping flags and infers store owner and group using them.grootfs delete-store
can delete a store, without deleting the backing filesystem.
0.15.0
- Do not leak corrupted volumes when GrootFS is killed half-way during a
grootfs create
. - Uses a more user-friendly error in
grootfs delete
when the image is not found. - Apply a consistent configuration file options naming convention.
- Remove the need of passing
--xfs-progs
togrootfs stats
in theoverlay-xfs
driver. - Allow newer layers to overwrite existing files with symlinks.
- Always create the whiteout device when using the
overlay-xfs
driver.
Fix some regressions and cf-deployment-ready
Docker support in overlay-xfs driver
- Stats doesn't fail when using overlay+xfs with no limits
- Fix whiteout implementation for overlay+xfs driver
- Support inclusive quotas for Overlay-XFS
The overlay-xfs filesystem driver can be used in Cloud Foundry
- Only with tar-based images at the moment.
v0.11.0 - Basic support for the overlay-xfs filesystem driver
Change image internal dirs permissions to 0755 * Users != store owner were unable to run any process in the container due to wrong permissions (was 0700) [#139824231] Signed-off-by: Claudia Beresford <cberesford@pivotal.io>
0.10.0
Fix test compilation after conflict [#138265057]
0.9.0
- Temporary blob files are now created inside the store (instead of the
$TEMP
directory).