This release fixes a few critical bugs in v0.22:
* osd: fix hang during mkfs journal creation
* objecter: fix rare hang during shutdown
* msgr: fix reconnect/disconnect errors due to timeouts
* init-ceph: check for correct instance in daemon_is_running()
* filestore: deliberate crash on ENOSPC or EIO to avoid corruption
* filestore: split xattrs into chunk (partial workaround for xattr size
limitations)
* filestore: ignore ENOSPC on xattrs until workaround for extN is
implemented
* radosgw: return 204 on removal of bucket/object
The main issue was the messenger bug, which causes internal cluster
messages to be dropped and triggered random cosd crashes. The ENOSPC/EIO
issue is also a bit strange: previously we silently ignored these errors.
Now we crash to avoid corrupting our data with partially committing
transaction, except in the case where we are trying to set a large xattr.
The problem here is that both btrfs have size limitations we hadn't
noticed before, btrfs on an individual xattr, and extN on the total size
of all xattrs. We will continue to ignore errors on extN (aside from
logging them) until we implement a workaround.
Relevant URLs:
* Direct download at: http://ceph.newdream.net/download/ceph-0.22.1.tar.gz
* For Debian packages, see http://ceph.newdream.net/wiki/Debian
sage
--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
