Merge "API Documentation for Monitoring Framework"

doc/source/devref/monitor-api.rst

@@ -0,0 +1,124 @@
+Tacker Monitoring Framework
+============================
+
+This section will introduce tacker monitoring framework and describes the
+various actions that a user can take when a specific event occurs.
+
+ - Introduction
+ - How to write a new monitor driver
+ - Events
+ - Actions
+ - How to write TOSCA template to monitor VNF entities
+
+Introduction
+-------------
+
+Tacker monitoring framework provides the NFV operators and VNF vendors to
+write a pluggable driver that monitors the various status coditions of the
+VNF entities it deploys and manages.
+
+How to write a new monitor driver
+----------------------------------
+
+A monitor driver for tacker is a python module which contains a class that inherits
+from "tacker.vm.monitor_drivers.abstract_driver.VNFMonitorAbstractDriver".
+If the driver depends/imports more than one module, then create a new python
+package under tacker/vm/monitor_drivers folder. After this we have to mention
+our driver path in setup.cfg file in root directory.
+
+For example:
+tacker.servicevm.monitor_drivers =
+    ping = tacker.vm.monitor_drivers.ping.ping:VNFMonitorPing
+
+The methods that need to override in driver are:
+1)def get_type(self)
+This method must return the type of driver ex: ping
+
+2)def get_name(self)
+This method must return the symbolic name of the device monitor plugin
+
+3)def get_description(self)
+This method must return the description for the monitor driver
+
+4)def monitor_get_config(self, plugin, context, device)
+This method must return dict of monitor configuration data
+
+5)def monitor_url(self, plugin, context, device)
+This method must return the url of device to monitor
+
+6)def monitor_call(self, device, kwargs)
+This method must either return boolean value True if VNF is healthy or return
+a event string like 'failure' or 'calls-capacity-reached' for specific VNF health
+condition.
+
+Custom events
+--------------
+As mentioned above, the return value of monitor_call is health status. If the
+return value of monitor_call is other than boolean value True, then we have to
+map those event to the corresponding action as described below.
+
+For example:
+    vdu1:
+      monitoring_policy:
+        ping:
+          actions:
+            failure: respawn
+In this  example we have an event called 'failure'. So whenever monitor_call returns
+'failure' tacker will respawn the VNF.
+
+
+Actions
+--------
+The available actions that a monitor driver can call when a particular event
+occurs
+
+1)respawn
+2)log_and_kill
+3)log
+
+How to write TOSCA template to monitor VNF entities
+----------------------------------------------------
+
+In the vdus section, under vdu you can specify the monitors details with
+corresponding actions and parameters.The syntax for writing monitor policy
+is as follows:
+
+vduN:
+  monitoring_policy:
+    <monitoring-driver-name>:
+      monitoring_params:
+        <param-name>: <param-value>
+        ...
+      actions:
+        <event>: <action-name>
+        ...
+    ...
+
+
+Example Template
+----------------
+::
+
+vdu1:
+  monitoring_policy:
+    ping:
+      actions:
+        failure: respawn
+
+vdu2:
+  monitoring_policy:
+    http-ping:
+      monitoring_params:
+        port: 8080
+        url: ping.cgi
+      actions:
+        failure: respawn
+
+    acme_scaling_driver:
+      monitoring_params:
+        resource: cpu
+        threshold: 10000
+      actions:
+        max_foo_reached: scale_up
+        min_foo_reached: scale_down
+