Shashank,
I will just show you how to get unique object id in cephfs, this maybe kind of
help to you in rgw.
I mount the kernel client in /cephdir , and version is ceph-v0.80.
1. list inodes of all files (inode is decimal)
root@ceph0:~# ls -i /cephdir
1099511627784 ceph-configure.sh 1099511627776 uuid 1099511627780 xx
decimal ----> hexdecimal
1099511627776 100000000000
1099511627780 100000000004
1099511627784 100000000008
2. list all objects in data (in hexdecimal)
for cephfs, all file objects are in pool data
root@ceph0:~# rados -p data ls -
10000000000.00000000
10000000008.00000000
10000000004.00000000
object id format is :
<inode_no>.<object_no>
If file more than size of object(usually 4194304 or 4MB),
file will be split into pieces of objects,
object id likes this:
10000000000.00000000, 10000000000.00000001, 10000000000.00000002 , ......
file oid mapping
object -> file
10000000000.00000000 -> uuid
10000000008.00000000 -> ceph-configure.sh
10000000004.00000000 -> xx
3. to comfirm file content
take /cephdir/uuid for example
get content of file:
root@ceph0:~# cat /cephdir/uuid
befbb574-e1a9-11e3-afed-a0481c7c0900
get content of object:
root@ceph0:~# rados -p data get 10000000000.00000000 -
befbb574-e1a9-11e3-afed-a0481c7c0900
object 10000000000.00000000 's content is the same as file /cephdir/uuid,
so, 10000000000.00000000 is the unique id of /cephdir/uuid (its size is 37 <
4914304, so there is only a object)
rgw 's data pool maybe the same as cephfs
min chen
_______________________________________________
ceph-users mailing list
ceph-users@lists.ceph.com
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
