Monitoring in CloudStack

Monitoring in Cloudstack
#CSCollab2022
“If it’s not monitored does it even exist?”
- Benjamin Franklin

2
$ whoami
Lucian-Paul Burlacu – lucian.burlacu@shapeblue.com
Long time sysadmin and penguinista
Cloudstack since 2013
Shapeblue since 2021

3
What is monitoring?
●
Periodic checks on progress or state coupled with suitable alerts.

5
What are people using?
●
There are many solid options for monitoring, but a community poll revealed
that Zabbix and Prometheus are the most popular.

6
Zabbix and Prometheus, a tale of old and new
Why Prometheus?
●
Free (Apache License) and gratis
●
Fast
●
Great query language – PromQL
●
Efficient storage, good performance
●
Simple to install
●
Great variety of “exporters”, easier to use than SNMP

7
Zabbix and Prometheus, a tale of old and new
Why Zabbix?
●
Free (GPLv2) and gratis
●
Simple to install, “yum/apt install” and some basic config
●
Nice UI with graph and history capability, API available
●
Administration is done via UI/API, no config files to mess with
●
“Crazy” flexible and configurable, could monitor anything, custom
commands, discovery
●
Powerful tiered alerting and actions + PROXY capability to get through VRs
●
Support out of the box for a great many technologies and devices (network
sw, IPMI, SNMP, routers, storage, vcenter etc)
●
“Speaks” Prometheus exporter

8
Why combine the 2 solutions?
●
The combined features and capabilities of Prometheus and Zabbix enable us
to deploy an almost complete solution for monitoring Cloudstack and the
entire software and hardware ecosystem around it.
●
Cons? Neither are simple, nor quick to get. There’s a learning curve
involved, as with most things.

10
Screenshot – Prometheus server

11
Let’s get into it – monitoring Cloudstack MGMT
●
Monitoring the Cloudstack management server with Zabbix can be done like
this:
- general Linux system info via either zabbix-agent, SNMP or Prometheus
exporter (zabbix-agent recommended) or a combination thereof.
- information on the Java internals, Zabbix supports this natively via the
Generic JMX (Java Management Extensions) template. Examples: Loaded
class count, Heap memory committed, Thread count and many others.

12
Let’s get into it – monitoring Cloudstack via JMX
(JMX Zabbix items screenshot)

13
Let’s get into it – monitoring Cloudstack via JMX
●
To enable JMX we need to edit /etc/default/cloudstack-management and add
the required parameters, YMMV:
Dcom.sun.management.jmxremote=true
Dcom.sun.management.jmxremote.port=3344
Dcom.sun.management.jmxremote.ssl=false
Dcom.sun.management.jmxremote.authenticate=false
●
You also need to (yum/apt) install zabbix-java-gateway package on the
Zabbix server and configure JavaGateway in zabbix_server.conf otherwise
Zabbix won’t be able to connect to JMX

14
Let’s get into it – monitoring Cloudstack MGMT
●
Monitoring Cloudstack via JMX

15
Monitoring Cloudstack via JMX

16
Let’s get into it
●
Moving on to Prometheus

17
Let’s get into it
●
Configuring Cloudstack’s Prometheus exporter:

19
Using the Cloudstack exporter in Zabbix
●
Zabbix does not talk to the Prometheus server, it retrieves the data from the
exporters directly.
●
You can of course deploy the Prometheus server, but Zabbix does not use it.

20
●
To add exported data in Zabbix we need to make sure there is an agent
connection registered for the host, then add a master item entry.

21
●
The next step is to add dependent items that actually parse the exported
data and add it into Zabbix, here’s an example:

22
●
Once this is done, we can add triggers to suit our needs.
●
There is some manual work involved, but it can be simplified by employing
Zabbix auto discovery (low level discovery)
-------------------------------------------------------
DEMO TIME on Cloudstack VMs triggers

23
Monitoring VMs behind Virtual Routers with Zabbix
Proxy
●
One of the great advantages of Zabbix for Cloudstack users is the Proxy
component. As the name suggests, it is a middleman between the Zabbix
server and the target VM.
●
Not only this, but the Proxy can work in an active mode, ie it connects to the
Zabbix server, not viceversa – this is crucial, because the opposite would
never work since the VR is locked down on the network.
●
Zabbix Proxy is lightweight and requires minimal configuration. It can use a
local or remote database, in our example it uses sqlite which required zero
work to set up.
●
Easy to install: apt update; apt install zabbix-proxy

24
Monitoring Vms behind Virtual Routers with Zabbix
Proxy

25
Proxy
Zabbix
Server
RX
Zabbix
Proxy running on the
Virtual Router*
(active)
Virtual
Machine
RX/TX
* If you do NOT have operator level access to the cloud and thus unable to work on the VR, then a
Zabbix Proxy could be installed in a new or existing VM in the same network with your target machines.
Alternatively, you can skip the Proxy component altogether if you can install Zabbix Agent and set it to
Active Mode in the target machine.

26
Proxy
DEMO TIME – Monitoring VMs behind Virtual Routers

27
Zabbix actions and alerting
●
Zabbix has a nice and powerful feature called “Actions”, as the name
suggests it takes various actions based on triggers, such as a VM no longer
responding to ICMP ping.
●
Based on this we can setup useful series of alerts and actions – such as
custom commands - to take place automatically:

28
Zabbix actions and alerting - self-healing
DEMO TIME – VM “self-healing” based on Zabbix Actions

29
Questions?
Any questions welcome.
Thank you!

30
Bonus slide: Prometheus libvirt exporter
Not out of time yet? DEMO TIME – Prometheus libvirt exporter!

Monitoring in CloudStack

More Related Content

What's hot

What's hot (20)

Similar to Monitoring in CloudStack

Similar to Monitoring in CloudStack (20)

More from ShapeBlue

More from ShapeBlue (20)

Recently uploaded

Recently uploaded (20)

Monitoring in CloudStack