docs/testing/user/userguide/opnfv_yardstick_tc058.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 TC058
   9 *************************************
  10
  11 +-----------------------------------------------------------------------------+
  12 |OpenStack Controller Virtual Router Service High Availability                |
  13 +==============+==============================================================+
  14 |test case id  | OPNFV_YARDSTICK_TC058:OpenStack Controller Virtual Router    |
  15 |              | Service High Availability                                    |
  16 +--------------+--------------------------------------------------------------+
  17 |test purpose  | This test case will verify the high availability of virtual  |
  18 |              | routers(L3 agent) on controller node. When a virtual router  |
  19 |              | service on a specified controller node is shut down, this    |
  20 |              | test case will check whether the network of virtual machines |
  21 |              | will be affected, and whether the attacked virtual router    |
  22 |              | service will be recovered.                                   |
  23 +--------------+--------------------------------------------------------------+
  24 |test method   | This test case kills the processes of virtual router service |
  25 |              | (l3-agent) on a selected controller node(the node holds the  |
  26 |              | active l3-agent), then checks whether the network routing    |
  27 |              | of virtual machines is OK and whether the killed service     |
  28 |              | will be recovered.                                           |
  29 +--------------+--------------------------------------------------------------+
  30 |attackers     | In this test case, an attacker called "kill-process" is      |
  31 |              | needed. This attacker includes three parameters:             |
  32 |              | 1) fault_type: which is used for finding the attacker's      |
  33 |              | scripts. It should be always set to "kill-process" in this   |
  34 |              | test case.                                                   |
  35 |              | 2) process_name: which is the process name of the load       |
  36 |              | balance service. If there are multiple processes use the     |
  37 |              | same name on the host, all of them are killed by this        |
  38 |              | attacker.                                                    |
  39 |              | 3) host: which is the name of a control node being attacked. |
  40 |              |                                                              |
  41 |              | In this case, this process name should set to "l3agent" ,    |
  42 |              | for example                                                  |
  43 |              | -fault_type: "kill-process"                                  |
  44 |              | -process_name: "l3agent"                                     |
  45 |              | -host: node1                                                 |
  46 +--------------+--------------------------------------------------------------+
  47 |monitors      | In this test case, two kinds of monitor are needed:          |
  48 |              | 1. the "ip_status" monitor that pings a specific ip to check |
  49 |              | the connectivity of this ip, which needs two parameters:     |
  50 |              | 1) monitor_type: which is used for finding the monitor class |
  51 |              | and related scripts. It should be always set to "ip_status"  |
  52 |              | for this monitor.                                            |
  53 |              | 2) ip_address: The ip to be pinged. In this case, ip_address |
  54 |              | will be either an ip address of external network or an ip    |
  55 |              | address of a virtual machine.                                |
  56 |              | 3) host: The node on which ping will be executed, in this    |
  57 |              | case the host will be a virtual machine.                     |
  58 |              |                                                              |
  59 |              | 2. the "process" monitor check whether a process is running  |
  60 |              | on a specific node, which needs three parameters:            |
  61 |              | 1) monitor_type: which used for finding the monitor class    |
  62 |              | and related scripts. It should be always set to "process"    |
  63 |              | for this monitor.                                            |
  64 |              | 2) process_name: which is the process name for monitor. In   |
  65 |              | this case, the process-name of monitor2 should be "l3agent"  |
  66 |              | 3) host: which is the name of the node running the process   |
  67 |              |                                                              |
  68 |              | e.g.                                                         |
  69 |              | monitor1-1:                                                  |
  70 |              | -monitor_type: "ip_status"                                   |
  71 |              | -host: 172.16.0.11                                           |
  72 |              | -ip_address: 172.16.1.11                                     |
  73 |              | monitor1-2:                                                  |
  74 |              | -monitor_type: "ip_status"                                   |
  75 |              | -host: 172.16.0.11                                           |
  76 |              | -ip_address: 8.8.8.8                                         |
  77 |              | monitor2:                                                    |
  78 |              | -monitor_type: "process"                                     |
  79 |              | -process_name: "l3agent"                                     |
  80 |              | -host: node1                                                 |
  81 +--------------+--------------------------------------------------------------+
  82 |metrics       | In this test case, there are two metrics:                    |
  83 |              | 1)service_outage_time: which indicates the maximum outage    |
  84 |              | time (seconds) of the specified Openstack command request.   |
  85 |              | 2)process_recover_time: which indicates the maximum time     |
  86 |              | (seconds) from the process being killed to recovered         |
  87 +--------------+--------------------------------------------------------------+
  88 |test tool     | None. Self-developed.                                        |
  89 +--------------+--------------------------------------------------------------+
  90 |references    | ETSI NFV REL001                                              |
  91 +--------------+--------------------------------------------------------------+
  92 |configuration | This test case needs two configuration files:                |
  93 |              | 1) test case file: opnfv_yardstick_tc058.yaml                |
  94 |              | -Attackers: see above "attackers" description                |
  95 |              | -Monitors: see above "monitors" description                  |
  96 |              | -Steps: the test case execution step, see "test sequence"    |
  97 |              | description below                                            |
  98 |              |                                                              |
  99 |              | 2)POD file: pod.yaml                                         |
 100 |              | The POD configuration should record on pod.yaml first.       |
 101 |              | the "host" item in this test case will use the node name in  |
 102 |              | the pod.yaml.                                                |
 103 +--------------+------+----------------------------------+--------------------+
 104 |test sequence | description and expected result                              |
 105 |              |                                                              |
 106 +--------------+--------------------------------------------------------------+
 107 |pre-test      | The test case image needs to be installed into Glance        |
 108 |conditions    | with cachestat included in the image.                        |
 109 |              |                                                              |
 110 +--------------+--------------------------------------------------------------+
 111 |step 1        | Two host VMs are booted, these two hosts are in two different|
 112 |              | networks, the networks are connected by a virtual router     |
 113 |              |                                                              |
 114 +--------------+--------------------------------------------------------------+
 115 |step 1        | start monitors:                                              |
 116 |              | each monitor will run with independently process             |
 117 |              |                                                              |
 118 |              | Result: The monitor info will be collected.                  |
 119 |              |                                                              |
 120 +--------------+--------------------------------------------------------------+
 121 |step 2        | do attacker: connect the host through SSH, and then execute  |
 122 |              | the kill process script with param value specified by        |
 123 |              | "process_name"                                               |
 124 |              |                                                              |
 125 |              | Result: Process will be killed.                              |
 126 |              |                                                              |
 127 +--------------+--------------------------------------------------------------+
 128 |step 4        | stop monitors after a period of time specified by            |
 129 |              | "waiting_time"                                               |
 130 |              |                                                              |
 131 |              | Result: The monitor info will be aggregated.                 |
 132 |              |                                                              |
 133 +--------------+--------------------------------------------------------------+
 134 |step 5        | verify the SLA                                               |
 135 |              |                                                              |
 136 |              | Result: The test case is passed or not.                      |
 137 |              |                                                              |
 138 +--------------+------+----------------------------------+--------------------+
 139 |post-action   | It is the action when the test cases exist. It will check    |
 140 |              | the status of the specified process on the host, and restart |
 141 |              | the process if it is not running for next test cases.        |
 142 |              | Virtual machines and network created in the test case will   |
 143 |              | be destoryed.                                                |
 144 |              |                                                              |
 145 +--------------+------+----------------------------------+--------------------+
 146 |test verdict  | Fails only if SLA is not passed, or if there is a test case  |
 147 |              | execution problem.                                           |
 148 +--------------+--------------------------------------------------------------+