On Wed, Jun 30, 2021 at 12:42 AM Manuel Holtgrewe <zyklenfrei@xxxxxxxxx> wrote: > > Dear all, > > I'm sorry if I'm asking for the obvious or missing a previous discussion of > this but I could not find the answer to my question online. I'd be happy to > be pointed to the right direction only. > > The cephfs-mirror tool in pacific looks extremely promising. How does it > work exactly? Is it based on files and (recursive) ctime or rather based on > object information? Does it handle incremental changes (only) between > snapshots? We started out with a design to use recursive stats (rctime) to identify changes in a directory tree, but soon hit a dead-end with bugs (in recursive stats, especially for snapshots) that were not straightforward to fix. Milind (cc) has made some good progress on fixing those. So, the mirror daemon design progressed with identifying changes between snapshots based on file mtime. This means a full walk on a directory tree to identify differences, which is not-so-performant for huge data sets. Moreover, this incremental mechanism is used between the current snapshot (to-be-syned) and the previous snapshot -- since the previous snapshot contents are present on the remote file system (for a directory root), the mirror daemon can do local snapshot comparison to identify changes between snapshots. > > There is an issue related to this that mentions recursive ctime. But that > would mean that users could "rsync -a" data to the file system and this > would not get synchronized. Do you mean the rctime not being synchronized? > > I have good experience with ZFS which is able to identify changes between > two snapshots A and B and then only transfer these changes (using a > sub-file level, on the ZFS equivalent of blocks to my understanding) to > another server with the same file system that is in the exact state as > snapshot A. Does cephfs-mirror work the same? Not really. The mirror daemon uses libcephfs to talk to local and remote file systems. There is no mirror daemon instance running on the remote cluster to "receive" data. > > Best wishes, > Manuel > _______________________________________________ > ceph-users mailing list -- ceph-users@xxxxxxx > To unsubscribe send an email to ceph-users-leave@xxxxxxx > -- Cheers, Venky _______________________________________________ ceph-users mailing list -- ceph-users@xxxxxxx To unsubscribe send an email to ceph-users-leave@xxxxxxx
 |