Cloud Storage with Red Hat Ceph Storage

You can use virtual extended attributes for information about CephFS file system use. The getfattr command, when used with the ceph attribute namespace on a directory, returns a list of recursive statistics for that directory.

[ceph: root@server /]# getfattr -d -m ceph.dir.* directory-path
file: directory-path
ceph.dir.entries="1"
ceph.dir.files="0"
ceph.dir.rbytes="424617209"
ceph.dir.rctime="1341146808.804098000"
ceph.dir.rentries="39623"
ceph.dir.rfiles="37362"
ceph.dir.rsubdirs="2261"
ceph.dir.subdirs="1"

The statistics provide detailed information.

Use cephfs set to enable snapshot creation for an existing CephFS file system.

[ceph: root@server /]# ceph fs set fs-name allow_new_snaps true

To create a snapshot, first mount the CephFS file system on your client node. Use the -o fs=_fs-name option to mount a CephFS file system when you have more than one. Then, create a subdirectory inside the .snap directory. The snapshot name is the new subdirectory name. This snapshot contains a copy of all the current files in the CephFS file system.

[root@node ~]# mount.ceph server.example.com:/ /mnt/mycephfs
[root@node ~]# mkdir /mnt/mycephfs/.snap/snap-name

Authorize the client to make snapshots for the CephFS file system with the s option.

[ceph: root@target /]# ceph fs authorize fs-name client path rws

To restore a file, copy it from the snapshot directory to another normal directory.

[root@node ~]# cp -a .snap/snap-name/file-name .

To fully restore a snapshot from the .snap directory tree, replace the normal entries with copies from the chosen snapshot.

[root@node ~]# rm -rf *
[root@node ~]# cp -a .snap/snap-name/* .

To discard a snapshot, remove the corresponding directory in .snap. The rmdir command succeeds even if the snapshot directory is not empty, without needing to use a recursive rm command.

[root@node ~]# rmdir .snap/snap-name

You can use CephFS to schedule snapshots. The snap_schedule module manages the scheduled snapshots. You can use this module to create and delete snapshot schedules. Snapshot schedule information is stored in the CephFS metadata pool.

To create a snapshot schedule, first enable the snap_schedule module on the MGR node.

[ceph: root@server /]# ceph mgr module enable snap_schedule

Then, add the new snapshot schedule.

[ceph: root@server /]# ceph fs snap-schedule add fs-path time-period [start-time]

If an earlier version than Python 3.7 is installed, then the start-time string must use the format %Y-%m-%dT%H:%M:%S. For Python version 3.7 or later, you can use more flexible date parsing. For example, to create a snapshot schedule to create a snapshot for the /volume folder every hour, you can use the ceph fs snap-schedule add command.

[ceph: root@server /]# ceph fs snap-schedule add /volume 1h
Schedule set for path /volume

On the client node, review the snapshots in the .snap folder on your mounted CephFS:

[root@node ~]# ls /mnt/mycephfs/.snap
scheduled-2021-10-06-08_00_00
scheduled-2021-10-06-09_00_00
scheduled-2021-10-06-10_00_00

You can list the snapshot schedules for a path with the list option:

[ceph: root@server /]# ceph fs snap-schedule list fs-path

Use the status option to verify the details for the snapshot schedules.

[ceph: root@server /]# ceph fs snap-schedule status fs-path

Remove a snapshot schedule by specifying the path.

[ceph: root@server /]# ceph fs snap-schedule remove fs-path

You can activate and deactivate snapshot schedules through the activate and deactivate options. When you add a snapshot schedule, it is activated by default if the path exists. However, if the path does not exist, then it is set as inactive, so you can activate it later when you create the path.

[ceph: root@server /]# ceph fs snap-schedule activate/deactivate