config.yaml description

clouds:
                     openstack: file:///<path_to_openrc>/openrc
                     openstack: user:passwd:tenant, http://......
                     ceph: <ssh_url>
                     fuel:
                                 url: http://<fuel_master_ip>:8000
                                 creds: <username><password>:<tenant_name>
                                 ssh_creds: <username>:<password>
                                 openstack_env: <enviroment_name>
                     ceph: local

 ssh_url: ssh://<username>@<host>::/<path_to_rsa_key>/id_rsa
 or ssh://<username>:<password>@<host>

Section clouds, describes clouds that are available and their credentials. Cloud type, as a key . Openstack cloud could be described as either like openrc location in file system or credentials to openstack cloud could be directly described in file. Fuel cloud has fixed structure with master node url, credentials both ssh and API and enviroment name. Ceph can be described in local or remote mode. Second one assume that full ssh url is provided and it allows you to connect ceph cloud. SSH url can be used with login/password or login/key file.

discover: fuel+openstack, ceph, openstack	

Discover section contains a list of clouds, whose nodes will be discovered on discover stage. It section contains list of clusters whose nodes should be discovered. Outcome of this stage is list of Node objects from nodes.node. Each node will have ssh url, list of roles taken from original cluster and connection.

explicit_nodes:
  "ssh://<username>@<host>:<path_to_key_file>": <node_name>
  "ssh://<username>:<password>@<host>": <node_name>

List of nodes, that should be used in testing, but aren't part of any cloud. User can use ssh urls either with login and password or login and key file.

logging:
extra_logs: 1	

Property responsible for printing extra logs. If yes set 1.

sensors:
   receiver_uri: udp://<ip>:<port>
   roles_mapping:
          ceph-osd: block-io #ceph-io, ceph-cpu, ceph-ram, ceph-net
          ceph-mon: ceph-io, ceph-cpu, ceph-ram, ceph-net
          os-compute: io, net
          test-vm: system-cpu
          <sensor_type>: <role_name>

Sensors that measures performance. Receiver url - where to send results of measurements. roles_mapping - mapping of node roles to sensor type. Receiver_uri can have explicit ip or {ip} in that case programm will automatically determina from which ip this machine is reachable.

tests: # $include(tests.yaml)
   start_test_nodes:
          openstack:
                 vm_params:
                        count: x1
                        img_name: disk_io_perf
                        flavor_name: disk_io_perf.256
                        keypair_name: disk_io_perf
                        network_zone_name: novanetwork
                        flt_ip_pool: nova
                        creds: "ssh://ubuntu@{0}::disk_io_perf.pem"
          internal_tests:
                 - pgbench:
                        opts:
                               num_clients: [4, 8, 12]
                               transactions: [1, 2, 3]            
                - io: 
                     tool: fio
                     config_file: tests/io_task_test.cfg
                - vm_count: 
                     max_lat_ms: 20
                     min_bw_mbps: 60
                     min_4k_direct_w_iops: 100
                     min_4k_direct_r_iops: 100

Test description, each test case has parameters of virtual machine that will be created for test and internal test section that has a list of tests and options for run that test on vm. You can either describe test in current config file or include another with test description.