go get github.com/ceph/go-ceph

The native RADOS library and development headers are expected to be installed.

Detailed documentation is available at https://godoc.org/github.com/ceph/go-ceph.

Connect to a Ceph cluster using a configuration file located in the default search paths.

conn, _ := rados.NewConn()
conn.ReadDefaultConfigFile()
conn.Connect()

A connection can be shutdown by calling the Shutdown method on the connection object (e.g. conn.Shutdown()). There are also other methods for configuring the connection. Specific configuration options can be set:

conn.SetConfigOption("log_file", "/dev/null")

and command line options can also be used using the ParseCmdLineArgs method.

args := []string{ "--mon-host", "1.1.1.1" }
err := conn.ParseCmdLineArgs(args)

For other configuration options see the full documentation.

Object in RADOS can be written to and read from with through an interface very similar to a standard file I/O interface:

// open a pool handle
ioctx, err := conn.OpenIOContext("mypool")

// write some data
bytes_in := []byte("input data")
err = ioctx.Write("obj", bytes_in, 0)

// read the data back out
bytes_out := make([]byte, len(bytes_in))
n_out, err := ioctx.Read("obj", bytes_out, 0)

if bytes_in != bytes_out {
    fmt.Println("Output is not input!")
}

The list of pools in a cluster can be retreived using the ListPools method on the connection object. On a new cluster the following code snippet:

pools, _ := conn.ListPools()
fmt.Println(pools)

will produce the output [data metadata rbd], along with any other pools that might exist in your cluster. Pools can also be created and destroyed. The following creates a new, empty pool with default settings.

conn.MakePool("new_pool")

Deleting a pool is also easy. Call DeletePool(name string) on a connection object to delete a pool with the given name. The following will delete the pool named new_pool and remove all of the pool's data.

conn.DeletePool("new_pool")

Contributions are welcome & greatly appreciated, every little bit helps. Make code changes via Github pull requests:

• Fork the repo and create a topic branch for every feature/fix. Avoid making changes directly on master branch.
• All incoming features should be accompanied with tests.
• Make sure that you run go fmt before submitting a change set. Alternatively the Makefile has a flag for this, so you can call make fmt as well.
• The integration tests can be run in a docker container, for this run:

make test-docker