Use librbd for rbd management in ceph-integration

[shtripat]

Currently ceph-integration executes rbd CLI commands for performing rbd management works. We should rather use librbd module for more controlled and organized way to handle rbd management.

Refer: http://docs.ceph.com/docs/master/rbd/librbdpy/ for more details.

[GowthamShanmugam]

@r0h4n @shtripat I have checked librbd, i have tested rbd creation it works fine, When is try to display rbd information using some built in function it gives info like

{'parent_name': '', 'parent_pool': 18446744073709551615L, 'num_objs': 256L, 'block_name_prefix': u'rbd_data.106d2ae8944a', 'obj_size': 4194304L, 'order': 22, 'size': 1073741824L}
it gives total size but it not give used size. I have trired with different functions but nothing is providing used information.

Current command is:
rbd du MyBlockDevice -p MyPool
NAME PROVISIONED USED
MyBlockDevice 1024M 0

here provisioning size is nothing but a total size of rbd, so librbd command gives szie 1073741824L it nothing but a 1gb. but no used size is present.

i filled some data with rbd then size become
[root@dhcp42-230 ~]# rbd du MyBlockDevice -p MyPool
NAME PROVISIONED USED
MyBlockDevice 1024M 4096k

but no change in rbd_inst.stat() output:
{'parent_name': '', 'parent_pool': 18446744073709551615L, 'num_objs': 256L, 'block_name_prefix': u'rbd_data.106d2ae8944a', 'obj_size': 4194304L, 'order': 22, 'size': 1073741824L}

And one more thing is, if cluster is in error state then rbd creation is hanging, no timeout for the create function so we have to timeout using some thread (import multiprocessing) https://codepaste.net/s4jczr.