docs/userguide/opnfv_yardstick_tc051.rst

   1 .. This work is licensed under a Creative Commons Attribution 4.0 International
   2 .. License.
   3 .. http://creativecommons.org/licenses/by/4.0
   4 .. (c) OPNFV, Yin Kanglin and others.
   5 .. 14_ykl@tongji.edu.cn
   6
   7 *************************************
   8 Yardstick Test Case Description TC051
   9 *************************************
  10
  11 +-----------------------------------------------------------------------------+
  12 |OpenStack Controller Node CPU Overload High Availability                     |
  13 |                                                                             |
  14 +--------------+--------------------------------------------------------------+
  15 |test case id  | OPNFV_YARDSTICK_TC051: OpenStack Controller Node CPU         |
  16 |              | Overload High Availability                                   |
  17 +--------------+--------------------------------------------------------------+
  18 |test purpose  | This test case will verify the high availability of control  |
  19 |              | node. When the CPU usage of a specified controller node is   |
  20 |              | stressed to 100%, which breaks down the Openstack services   |
  21 |              | on this node. These Openstack service should able to be      |
  22 |              | accessed by other controller nodes, and the services on      |
  23 |              | failed controller node should be isolated.                   |
  24 +--------------+--------------------------------------------------------------+
  25 |test method   | This test case stresses the CPU uasge of a specified control |
  26 |              | node to 100%, then checks whether all services provided by   |
  27 |              | the environment are OK with some monitor tools.              |
  28 +--------------+--------------------------------------------------------------+
  29 |attackers     | In this test case, an attacker called "stress-cpu" is        |
  30 |              | needed. This attacker includes two parameters:               |
  31 |              | 1) fault_type: which is used for finding the attacker's      |
  32 |              | scripts. It should be always set to "stress-cpu" in          |
  33 |              | this test case.                                              |
  34 |              | 2) host: which is the name of a control node being attacked. |
  35 |              | e.g.                                                         |
  36 |              | -fault_type: "stress-cpu"                                    |
  37 |              | -host: node1                                                 |
  38 +--------------+--------------------------------------------------------------+
  39 |monitors      | In this test case, the monitor named "openstack-cmd" is      |
  40 |              | needed. The monitor needs needs two parameters:              |
  41 |              | 1) monitor_type: which is used for finding the monitor class |
  42 |              | and related scritps. It should be always set to              |
  43 |              | "openstack-cmd" for this monitor.                            |
  44 |              | 2) command_name: which is the command name used for request  |
  45 |              |                                                              |
  46 |              | There are four instance of the "openstack-cmd" monitor:      |
  47 |              | monitor1:                                                    |
  48 |              | -monitor_type: "openstack-cmd"                               |
  49 |              | -command_name: "nova image-list"                             |
  50 |              | monitor2:                                                    |
  51 |              | -monitor_type: "openstack-cmd"                               |
  52 |              | -command_name: "neutron router-list"                         |
  53 |              | monitor3:                                                    |
  54 |              | -monitor_type: "openstack-cmd"                               |
  55 |              | -command_name: "heat stack-list"                             |
  56 |              | monitor4:                                                    |
  57 |              | -monitor_type: "openstack-cmd"                               |
  58 |              | -command_name: "cinder list"                                 |
  59 +--------------+--------------------------------------------------------------+
  60 |metrics       | In this test case, there is one metric:                      |
  61 |              | 1)service_outage_time: which indicates the maximum outage    |
  62 |              | time (seconds) of the specified Openstack command request.   |
  63 +--------------+--------------------------------------------------------------+
  64 |test tool     | Developed by the project. Please see folder:                 |
  65 |              | "yardstick/benchmark/scenarios/availability/ha_tools"        |
  66 |              |                                                              |
  67 +--------------+--------------------------------------------------------------+
  68 |references    | ETSI NFV REL001                                              |
  69 |              |                                                              |
  70 +--------------+--------------------------------------------------------------+
  71 |configuration | This test case needs two configuration files:                |
  72 |              | 1) test case file: opnfv_yardstick_tc051.yaml                |
  73 |              | -Attackers: see above "attackers" discription                |
  74 |              | -waiting_time: which is the time (seconds) from the process  |
  75 |              | being killed to stoping monitors the monitors                |
  76 |              | -Monitors: see above "monitors" discription                  |
  77 |              | -SLA: see above "metrics" discription                        |
  78 |              |                                                              |
  79 |              | 2)POD file: pod.yaml                                         |
  80 |              | The POD configuration should record on pod.yaml first.       |
  81 |              | the "host" item in this test case will use the node name in  |
  82 |              | the pod.yaml.                                                |
  83 |              |                                                              |
  84 +--------------+--------------------------------------------------------------+
  85 |test sequence | description and expected result                              |
  86 |              |                                                              |
  87 +--------------+--------------------------------------------------------------+
  88 |step 1        | start monitors:                                              |
  89 |              | each monitor will run with independently process             |
  90 |              |                                                              |
  91 |              | Result: The monitor info will be collected.                  |
  92 |              |                                                              |
  93 +--------------+--------------------------------------------------------------+
  94 |step 2        | do attacker: connect the host through SSH, and then execute  |
  95 |              | the stress cpu script on the host.                           |
  96 |              |                                                              |
  97 |              | Result: The CPU usage of the host will be stressed to 100%.  |
  98 |              |                                                              |
  99 +--------------+--------------------------------------------------------------+
 100 |step 3        | stop monitors after a period of time specified by            |
 101 |              | "waiting_time"                                               |
 102 |              |                                                              |
 103 |              | Result: The monitor info will be aggregated.                 |
 104 |              |                                                              |
 105 +--------------+--------------------------------------------------------------+
 106 |step 4        | verify the SLA                                               |
 107 |              |                                                              |
 108 |              | Result: The test case is passed or not.                      |
 109 |              |                                                              |
 110 +--------------+--------------------------------------------------------------+
 111 |post-action   | It is the action when the test cases exist. It kills the     |
 112 |              | process that stresses the CPU usage.                         |
 113 +--------------+--------------------------------------------------------------+
 114 |test verdict  | Fails only if SLA is not passed, or if there is a test case  |
 115 |              | execution problem.                                           |
 116 |              |                                                              |
 117 +--------------+--------------------------------------------------------------+