Aline, as posted inn my RFC patch: [project-kimchi][RFC]Storage backend targets probeOn 12/30/2013 12:17 AM, lvroyce@linux.vnet.ibm.com wrote:
From: Royce Lv <lvroyce@linux.vnet.ibm.com> v3>v4, fix inconsistency between doc and json schema v1>v3, fix racing problem, fix style. Add parameters to GET request so that we will query storage server as: /storageservers?type=netfs Royce Lv (5): Support params for GET method Add testcase for GET param Storage server: Update API.md storage server: update controller.py storage server: Update model and mockmodel docs/API.md | 13 +++++++++++++ src/kimchi/API.json | 11 +++++++++++ src/kimchi/controller.py | 26 ++++++++++++++++++++++++-- src/kimchi/mockmodel.py | 13 +++++++++++++ src/kimchi/model.py | 14 +++++++++++++- src/kimchi/root.py | 1 + tests/test_rest.py | 37 +++++++++++++++++++++++++++++++++++++ 7 files changed, 112 insertions(+), 3 deletions(-)
Let me explain what I expect with this patch set:
GET /storageservers will return all storage servers
Example:
[
{'host': 'localhost',
'target_type': 'netfs',
'sources': ['/srv', '/mnt/isos'],
}
{'host': '123.234.123.12',
'target_type': 'iscsi',
'sources': ['iqn.2013-06.com.example:iscsi-pool'],
}
]
Because a host can serve as multiple server(iscsi, gluster, nfs) at same time. So I choose to make user query a specific kind by:
GET /storageservers?target_type=<server-type> will return only the storage servers with type <server-type>
Example: GET /storageservers?target_type=netfs
[
{'host': 'localhost',
'target_type': 'netfs',
'sources': ['/srv', '/mnt/isos'],
}
]
GET /storageservers/<id> will return data related to this specific server
Example: GET /storageservers/localhost
{'host': 'localhost',
'target_type': 'netfs',
'sources': ['/srv', '/mnt/isos'],
}
The 'sources' for netfs, for example, will be the parsed output from "showmount -e <host>" command.I don't quite agree... 'POST' to storage server means you want to create a storage server.
So when the user want to create NFS pool with a new server (never used before)
POST /storageservers/new-nfs-server
{'host': 'new-nfs-server',
'target_type': 'netfs',
'sources': ['/srv', '/mnt'],
}
Then the user know which paths are available to create the NFS pool
Does that make sense for you?