(Copying ceph-users to share the info more broadly) On Thu, Nov 24, 2016 at 10:50 AM, <ren.huanwen@xxxxxxxxxx> wrote: > Hi John > > I have some questions about the use of cephfs, > Can you help me answer, Thank you! > > We built a Openstack(M) file share, and use Manila componets based CephFS. > I can export CephFS'posix and refer to > http://docs.openstack.org/developer/manila/devref/cephfs_native_driver.html(CephFS > Native driver from you, :+1:) > as we all know: > use "nfs-ganesha", I can manual export NFS based on CephFS, and not in > Manila; > use "samba 4.x", I can manual exprot CIFS based on CephFS, and not in > Manila. > but if i want to direct export the CIFS and NFS based on CephFS in Manila, > seems not support. > although Manila support Ganesha Library(not support Samba library) > so will plan to support the above functions? An NFS+CephFS driver for Manila is a work in progress. The rough plan is to have some initial functionality for auto-configuring ganesha exports using the existing ganesha modules in openstack Ocata, and then the to have it automatically creating gateway VMs in the subsequent Pike release. > Two other issues outside > 1. ceph-fuse cannot mount cephfs snapshot directory? > Example:ceph-fuse /root/mycephfs5 --id=admin --conf=./client.conf > --keyring=/etc/ceph/ceph.client.admin.keyring > --client-mountpoint=/volumes/_nogroup/b53cbff4-a3f2-402b-91c2-aaf967f32d40/.snap/ Hmm, we've never tested this, and it probably needs some work because snapshot dirs are special. This will be necessary to enable Manila's mountable snapshots feature: https://github.com/openstack/manila-specs/blob/master/specs/ocata/mountable-snapshots.rst I've created at ticket here: http://tracker.ceph.com/issues/18050, although I'm not sure how soon it would reach the top of anyone's priority list given the current focus on robustness and multi-mds. > 2. if i delete cephfs using "ceph fs rm {cephfs_name}", but there is > cephfs_data_pool and cephfs_meta_pool, > Based on the old cephfs_data_pool and cephfs_meta_pool, whether can > restore cephfs. This would be considered a disaster recovery situation. Removing the filesystem doesn't touch anything in the pools, so you'd do something like: * Stop all MDS daemons * Do a "ceph fs new" with the same pools * Do a "ceph fs reset" on the new filesystem to make it skip the creating stage. * If you had multiple active MDSs you would probably also need to do some extra work to truncate journals etc. You can also use cephfs-data-scan to (best effort) scrape files directly out of a cephfs data pool to some other filesystem. The usual warnings for disaster recovery tools all apply to the situation of trying to recover a cephfs filesystem from pools: this is a last resort, they can do harm as well as good, and anyone unsure should seek expert advice before using them. John _______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
 |