Python C Shell C++ Makefile PowerShell Other
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
.idea Add: new 0.9-beta1 version, with: Aug 23, 2016
appveyor Enh: TEST first windows test, with appveyor Jul 13, 2018
bash_completion Enh: rebranding done (ง︡’-‘︠)ง Aug 3, 2017
bin Enh: change print calls to print() ones, to prepare python3 compat. May 1, 2018
data Enh: PACKS split hypervisor into multiple packs Jul 26, 2018
doc Enh: more tests about Alpine linux Nov 13, 2017
etc Add: Topics can be now listed and enable/disable Mar 25, 2018
images Add: buzz play_tutorial command Oct 18, 2017
init.d Fix: INIT.D the stop did wait too long before exit May 9, 2018
opsbro Enh: move core packs to it's own level. If works, then make it read only Jul 23, 2018
spike Fix: more python3 stuff Jul 1, 2018
test Fix: TEST HTTP demo, switch DNS system query hook by the new standard… Aug 2, 2018
ui Add: now collectors & modules parameters are declared by the elements… Aug 8, 2017
.gitignore Add: python static build Oct 26, 2017
.travis.yml Enh: readme Jul 13, 2018
LICENSE 0.9 commit version Feb 25, 2015
MANIFEST.in Fix: python pip installation fix, was mmissing the init.d script in t… Sep 5, 2017
README.md Enh: readme Jul 13, 2018
appveyor.yml Add: window tests about shinken-enterprise module in a one shot run Jul 17, 2018
deploy.sh Add: header now show OpsBro 6 topics with links between thems (color … Sep 7, 2017
leanops.txt Add: agent start --one-shot to execute the agent not in a daemon form… Sep 27, 2017
opsbro.spec Fix: do not consider address that starts with 127. as allowed public IPs Oct 30, 2017
run-ui.sh Enh: [ui] add sample script for run ui. Nov 22, 2016
setup-py2exe.py Enh: prepare a python3 compat, bu changing all iteritems() to items. … May 13, 2018
setup.py Fix: windows: remove 3.4 version, and fir a python3 issue in other py… Jul 17, 2018
technical_spec_and_todo.txt Enh: remove leveldb dependency, by switching to sqlite if not installed. May 3, 2018

README.md

Build Status

OpsBro

Agent

All the configuration / deployment is model based (aka into packs)

Installation

Prerequites

You will need:

  • python (2.6, 2.7 or 3.4+ versions)

Installation

Just launch:

python setup.py install

Example of the installation at:

Note: main linux distributions are managed and will automatically install dependencies from the package manager (apt/yum/apk/zypper):

  • Alpine 3.1→3.7
  • Amazonlinux 2016 & 2017
  • Centos 6 & 7
  • Debian 6→9
  • Fedora 24→28
  • OpenSuse 42
  • Ubuntu 12.04→18.10

On others distributions depedencies will be take from pypi.

Start OpsBro daemon

You can start opsbro as a daemon with:

service opsbro start

Stop OpsBro daemon

Just launch:

service opsbro stop

Display OpsBro information

Just launch:

opsbro agent info

You will have several information about the current opsbro agent state:

Automatic detection (os, apps, location, ...)

Detectors are rules that are executed by the agent to detect your server properties like

  • OS (linux, redhat, centos, debian, windows, ...)
  • Applications (mongodb, redis, mysql, apache, ...)
  • Location (country, city, GPS Lat/Long)

You should declare a json object like:

detector:
    apply_if: "grep_file('centos', '/etc/redhat-release')"
    add_groups:
        - linux
        - centos
    interval: 3600s
  • If there is the string centos in the file /etc/redhat-release
  • Then add the group "linux" and centos" to the local agent
  • Execute every 3600 seconds

Service discovery

You can use OpsBro on one or more server in an independant way, but it's full power is when you are linking them together in a cluster.

Agent cluster membership

Add your local node to the node cluster

First you need to install and launch the node in another server.

Then in this other server you can launch:

opsbro gossip join  OTHER-IP
Auto discover LAN nodes (UDP broadcast detection)

If your nodes are on the same LAN, you can use the UDP auto-detection to list others nodes on your network. I will send an UDP broadcast packet that other nodes will answer.

NOTE: if you are using an encryption key (recommanded) then you must already have set it. If not, the other node won't answer to your query.

opsbro gossip detect --auto-join

Agent

List your opsbro cluster members

You can list the cluster members on all nodes with :

opsbro  gossip members

Agent

Metrology: collect your server metrics (cpu, kernel, databases metrics, etc)

Collectors are code executed by the agent to grok and store local os or application metrics.

You can list available collectors with the command:

opsbro collectors list

Agent

How to see collected data? (metrology)

The opsbro agent is by default getting lot of metrology data from your OS and applications. It's done by "collctors" objets. You can easily list them and look at the collected data by launching:

opsbro collectors show system

For example for the system collector:

Agent

Show dashboards

Packs can provide dashboards to show collected data. Here is an example of the standard linux dashboard:

opsbro dashboards show linux

Agent

To list available dashboards: opsbro dashboards list

Export and store your application telemetry into the agent metric system

Real time application performance metrics

The statsd protocol is a great way to extract performance statistics from your application into your monitoring system. You application will extract small timing metrics (like function execution time) and send it in a non blocking way (in UDP).

The statsd daemon part will agregate counters for 10s and will then export the min/max/average/99thpercentile to a graphite server so you can store and graph them.

In order to enable the statsd listener, you must define enable it your configuration.

You can show it with the command:

opsbro packs show

And search for the statsd pack:

Agent

All you need is to change the enabled parameter of the statsd pack.

You can get a description about the parameter first with:

opsbro packs parameters get global.statsd.enabled

And you can edit it with:

opsbro packs parameters set global.statsd.enabled true

Agent

Then restart our agent:

service opsbro restart

Grafana: view your node metrology data into Grafana

You can see your metrology data into Grafana. All you need to do is to enable the grafana pack for this.

Agent

You will need:

  • set a grafana API key

  • change the grafana URI if not on the same server as your agent

  • add the group grafana-connector to your agent so it will start the grafana module

    opsbro packs parameters set global.grafana.api_key 'YOUR-API-KEY' opsbro packs parameters set global.grafana.uri 'http://YOUR-GRAFANA-SERVER:3000'

And to add the group grafana-connector to your agent:

opsbro agent parameters add groups grafana-connector

And voila, you will see a new data-source in your grafana:

Agent

Monitoring

You can execute checks on your agent by two means:

  • Use the collectors data and evaluate check rule on it
  • Execute a nagios-like plugin

Common check parameters for evaluated and nagios plugins based checks

Some parameters are common on the two check types you can defined.

  • interval: how much seconds the checks will be scheduled
  • if_group: if present, will declare and execute the check only if the agent group is present

Evaluate check rule on collectors data

Evaluated check will use collectors data and should be defined with:

  • ok_output: python expression that create a string that will be shown to the user if the state is OK
  • critical_if: python expression that try to detect a CRITICAL state
  • critical_output: python expression that create a string that will be shown to the user if the state is CRITICAL
  • warning_if: python expression that try to detect WARNING state
  • warning_output: python expression that create a string that will be shown to the user if the state is WARNING
  • interval: launch the check every X seconds

The evaluation is done like this:

  • if the critical expression is True => CRITICAL
  • else if warning expression is True => WARNING
  • else go OK

For example here is a memory check on a linux server:

check:
    if_group: linux

    ok_output: "'OK: memory is at %d%%' % {{collector.memory.phys_used}}"

    critical_if: "{{collector.memory.phys_used}} > {{parameters.memory_critical}}"
    critical_output: "'CRITICAL: memory is at %d%%' % {{collectors.memory.phys_used}}"

    warning_if: "{{collector.memory.phys_used}} > {{parameters.threshold.memory_warning}}"
    warning_output:  "'WARNING: memory is at %d%%' % {{collector.memory.phys_used}}"

    interval: "30s"

You can have the result with the command:

opsbro monitoring state

Agent

Use Nagios plugins

Nagios based checks will use Nagios plugins and run them. Use them if you don't have access to the information you need in the collectors.

The parameter for this is:

  • script: the command line to execute your plugin

Here is an example

check:
    if_group: linux
    script:   "/var/lib64/nagios/check_mailq -w 1 -c 2"
    interval: 60s

Notify check/node state change with emails

You can be notified about check state changed with handlers. Currently 2 are managed:

  • email
  • slack

Email handlers

You can get a mail each time a check did change state. All is managed by the mail pack. You can enable it with:

opsbro packs parameters set global.mail.enabled true

Notify check & node state change into slack

You can be notified about check state changed with handlers:

Agent

You can enable it with:

opsbro agent parameters add groups slack
opsbro packs parameters set global.slack.token    'SLACK-TOKEN'

Export your nodes and check states into Shinken or Nagios

You can export all your nodes informations (new, deleted or change node) into your Shinken installation. It will automatically:

  • create new host when you start a new node
  • change the host configuration (host templates) when a new group is add/removed on your agent
  • remove your host when you delete your agent (by terminating your Cloud instance for example)

All is managed into the shinken pack.

For Nagios:

opsbro  packs parameters set local.shinken.enabled               true
opsbro  packs parameters set local.shinken.cfg_path              /usr/local/nagios/etc/objects/agent
opsbro  packs parameters set local.shinken.external_command_file /usr/local/nagios/var/rw/nagios.cmd
opsbro  packs parameters set local.shinken.reload_command        "/etc/init.d/nagios reload"
opsbro  packs parameters set local.shinken.monitoring_tool       nagios

Note that you must create the /usr/local/nagios/etc/objects/agent and declare it in the nagios.cfg file.

For Shinken:

opsbro  packs parameters set local.shinken.enabled               true
opsbro  packs parameters set local.shinken.cfg_path              /etc/shinken/agent
opsbro  packs parameters set local.shinken.external_command_file /var/lib/shinken/shinken.cmd
opsbro  packs parameters set local.shinken.reload_command        "/etc/init.d/shinken reload"
opsbro  packs parameters set local.shinken.monitoring_tool       shinken

For Shinken, note that the external unix pipe (/var/lib/shinken/shinken.cmd) is available with the receiver module named-pipe.

DNS: Access your nodes informations by DNS

If you enable the DNS interface for your agent, it will start an internal DNS server that will answer to queries. So your applications will be able to easily exchange with the valid node that is still alive or launched.

All you need is to set your agent to the dns-listener group to start the DNS listener module.

opsbro agent parameters add groups dns-listener

Then you can query your agent in DNS:

dig -p 6766  @127.0.0.1 linux.group.local.opsbro
192.168.56.103
192.168.56.105

It list all available node with the "group" linux.

Is there an UI available?

Soon :)