summaryrefslogtreecommitdiffstats
path: root/docs/development/manuals
diff options
context:
space:
mode:
authorUmar Farooq <umar.farooq@neclab.eu>2017-07-13 12:20:56 +0200
committerRyota Mibu <r-mibu@cq.jp.nec.com>2017-08-09 04:39:54 +0000
commit282369b6fd58a78e6a7c91f21b331363d4ed0fb3 (patch)
tree3dc95a2c16ca948627df6fb93629a3652e620211 /docs/development/manuals
parentdb6d9cdcb37fee810fdf59e4d96b9a5139b5c8b7 (diff)
Add Collectd as a Monitor Type
A plugin for collectd is added to use collectd on compute as a monitor type. Monitor files are updated accordingly. The inspector now listens on all interfaces instead of only localhost to enable it to communicate with compute node. JIRA: DOCTOR-86 JIRA: DOCTOR-101 Change-Id: Idc834d428152e4687020eff7d8db36a652b1bf86 Signed-off-by: Umar Farooq <umar.farooq@neclab.eu>
Diffstat (limited to 'docs/development/manuals')
-rw-r--r--docs/development/manuals/monitors.rst36
1 files changed, 36 insertions, 0 deletions
diff --git a/docs/development/manuals/monitors.rst b/docs/development/manuals/monitors.rst
new file mode 100644
index 00000000..0d22b1de
--- /dev/null
+++ b/docs/development/manuals/monitors.rst
@@ -0,0 +1,36 @@
+.. This work is licensed under a Creative Commons Attribution 4.0 International License.
+.. http://creativecommons.org/licenses/by/4.0
+
+Monitor Types and Limitations
+=============================
+
+Currently there are two monitor types supported: sample and collectd
+
+Sample Monitor
+--------------
+
+Sample monitor type pings the compute host from the control host and calculates the
+notification time after the ping timeout.
+Also if inspector type is sample, the compute node needs to communicate with the control
+node on port 12345. This port needs to be opened for incomming traffic on control node.
+
+Collectd Monitor
+----------------
+
+Collectd monitor type uses collectd daemon running ovs_events plugin. Collectd runs on
+compute to send instant notification to the control node. The notification time is
+calculated by using the difference of time at which compute node sends notification to
+control node and the time at which consumer is notified. The time on control and compute
+node has to be synchronized for this reason. For further details on setting up collectd
+on the compute node, use the following link:
+http://docs.opnfv.org/en/stable-danube/submodules/barometer/docs/release/userguide/feature.userguide.html#id18
+
+Collectd monitors an interface managed by OVS. If the interface is not be assigned
+an IP, the user has to provide the name of interface to be monitored. The command to
+launch the doctor test in that case is:
+MONITOR_TYPE=collectd INSPECTOR_TYPE=sample INTERFACE_NAME=example_iface ./run.sh
+
+If the interface name or IP is not provided, the collectd monitor type will monitor the
+default management interface. This may result in the failure of doctor run.sh test case.
+The test case sets the monitored interface down and if the inspector (sample or congress)
+is running on the same subnet, collectd monitor will not be able to communicate with it.