nagios-plugins-onedata

This repository contains Nagios plugins for testing Onedata services availability, namely Oneprovider and Onezone services.

Requirements

These Nagios probes have the following requirements in order to work properly:

Bash
cURL
xmllint (from libxml2-utils)
access to port 6666 on machine hosting Onedata service

Running the probes

Oneprovider

Oneprovider probe supports the following command line options:

Nagios plugin for Oneprovider service of Onedata

Usage:
src/check_oneprovider [-h|--help] [-t|--timeout <timeout>] [-p|--port <port>]
                    [-H|--hostname <hostname>] [-u|--url <endpoint>]

 -h|--help    - Prints this help
 -t|--timeout - Determines the entire connection timeout (Default: 10 seconds)
 -H|--host    - The Onedata hostname
 -p|--port    - The Onedata Nagios port (Default: 6666)
 -u|--url     - The complete Onedata Nagios endpoint

Onezone

Nagios plugin for Onezone service of Onedata

Usage:
src/check_oneprovider [-h|--help] [-t|--timeout <timeout>] [-p|--port <port>]
                    [-H|--hostname <hostname>] [-u|--url <endpoint>]

 -h|--help    - Prints this help
 -t|--timeout - Determines the entire connection timeout (Default: 10 seconds)
 -H|--host    - The Onedata hostname
 -p|--port    - The Onedata Nagios port (Default: 6666)
 -u|--url     - The complete Onedata Nagios endpoint

How the probes work

Each of Onedata services (Onezone, Oneprovider) expose an endpoint with health data in XML for all worker processes running on a given site (each Onedata service can have mutliple worker processes running in parallel for higher performance and reliability).

By default all Onedata service provide this information on port 6666 so it must be ensured that this port is opened on the Onedata sites to the nodes where Nagios monitoring probes are executed.

An example XML output for a Onezone service is below:

<?xml version="1.0"?>
<healthdata date="2016/11/24 13:10:02" status="error">
  <oz_worker name="oz_worker@node1.onezone.localhost" status="ok">
    <node_manager status="ok"/>
    <request_dispatcher status="ok"/>
    <changes_worker status="ok"/>
    <datastore_worker status="ok"/>
    <dns_worker status="ok"/>
    <groups_graph_caches_worker status="ok"/>
    <ozpca_worker status="ok"/>
    <subscriptions_worker status="ok"/>
    <dns_listener status="ok"/>
    <gui_listener status="ok"/>
    <nagios_listener status="ok"/>
    <oz_redirector_listener status="ok"/>
    <rest_listener status="ok"/>
    <subscriptions_wss_listener status="ok"/>
  </oz_worker>
  <oz_worker name="oz_worker@node2.onezone.localhost" status="error">
    <node_manager status="ok"/>
    <request_dispatcher status="ok"/>
    <changes_worker status="ok"/>
    <datastore_worker status="ok"/>
    <dns_worker status="ok"/>
    <groups_graph_caches_worker status="error"/>
    <ozpca_worker status="ok"/>
    <subscriptions_worker status="ok"/>
    <dns_listener status="ok"/>
    <gui_listener status="ok"/>
    <nagios_listener status="ok"/>
    <oz_redirector_listener status="error"/>
    <rest_listener status="ok"/>
    <subscriptions_wss_listener status="ok"/>
  </oz_worker>
</healthdata>

In case all workers are working and the healthdata status is ok the probe will return OK status.

If all of the workers are down and return error status the probe will return a CRITICAL status code and in case at least one worker is still running the status will be WARNING.

Building CentOS RPM

Building can be done on a Centos 6 Docker container:

#
# Build the RPM from scratch 
#
$ docker run -ti centos:6 /bin/bash
$ yum install git rpm-build
$ cd /root
$ mkdir -p ~/rpmbuild/{BUILD,RPMS,SOURCES,SPECS,SRPMS}
$ git clone https://github.com/onedata/nagios-plugins-onedata
$ cd nagios-plugins-onedata
$ make
$ cp nagios-plugins-onedata*tar.gz ../rpmbuild/SOURCES
$ rpmbuild -bb nagios-plugins-onedata.spec

#
# Install and test the RPM
#
$ cd ../rpmbuild/RPMS/noarch
$ rpm -Uvh nagios-plugins-onedata-3.0.0-1.el6.noarch.rpm
$ /usr/lib64/argo-monitoring/probes/org.onedata/check_onezone -h

Running probe unit tests

First create and start the Docker, which serves the mockup healthdata XML files:

$ cd nagios-plugins-onedata
$ docker build --file docker/Dockerfile.nginx -t onedata-nagios-test .
$ docker run -t -d -p 80:80 onedata-nagios-test

Then run the tests/test.sh script:

$ tests/test.sh

Finally stop the Nginx container and optionally destroy it:

$ docker stop $(docker ps -f ancestor=onedata-nagios-test -q)
$ docker rmi onedata-nagios-test

eimamagi/nagios-plugins-onedata