Pre-requisites : Centreon server¶
Check that the parameter date.timezone is correctly configured in /etc/opt/rh/rh-php71/php.ini (same timezone displayed with the command “timedatectl status” )
Avoid the usage of the following variables in your monitoring MySQL configuration. They stop long queries execution and can stop the ETL or the report generation jobs:
To capitalize on the application availability reporting
Users and groups
Description of users, umask and home directory
Pre-requisites: reporting server¶
|< 4 000||2 CPU ( 3Ghz ) minimum||12Go minimum|
|< 20 000||4 CPU (3GHz) minimum||16Go minimum|
|>= 20 000 and < 40 000||4 CPU (3GHz) minimum||24Go minimum|
|> 40 000 and < 100 000||8 CPU (3GHz) minimum||32Go minimum|
|> 100 000||Contact Centreon|
Storage space: use the following storage estimation file
|/var (containing MySQL data)||Use the result of the excel file in the table above|
|MySQL temp folder||highly recommanded to define the tmpdir of MySQL on /var to prevent /tmp to get full|
|Volume group*||5GB minimum of free space on the Volum group hosting the data storage MySQL/MariaDB DBMS|
To control the free space use the command below replacing vg_data by the Volum group name:
vgdisplay vg_data | grep -i free
|OS||CentOS 7 / Redhat 7|
Dependencies automatically installed :
|Java||Openjdk >= 1.7|
We advise to tune your MySQL database server on your reporting server in order to have better performance. You will need at least 12GB on your reporting server to run the configuration file provided below. Add the following file (centreon.cnf) on your reporting server in /etc/my.cnf.d/ by the following one: . Make sure to have a tmp folder in /var/lib/mysql.
Do not make these optimizations on your monitoring server.
Users and groups :
Description of users, umask and home directory:
In order to be able to use Centreon MBI, you need to retrieve a license file.
To generate it, we need the fingerprint of the machine on which Centreon MBI’s interface is installed (Central server). The fingerprint is generated using Centreon Fingerprint which is packaged in RPM and is available on the Centreon repository.
Install the Centreon Fingerprint binary.
# yum install centreon-fingerprint
Execute Centreon Fingerprint
The machine fingerprint will be displayed in the console.
You will have to provide it to Centreon Support that will generate the license file you need to use Centreon MBI. You’ll install it during the installation procedure.
Monitoring best practices for reporting¶
Plugins and data¶
To be able to have reporting on performance data using default Centreon MBI report , you have to monitor at least some basic performance indicators (metrics):
- CPU: Must return a percentage value, using one or multiple metrics and 100 as the maximum value (even if it’s obvious)
- Must return at least one metric
- The memory usage: the value must be in Bytes
- The memory usage warning threshold
- The memory usage critical threshold
- The total allocated memory un Bytes
The memory plugin used to monitor this indicator must return the following output:status information | metric_name=valueunit;warning_threshold;critical_threshold;min_value;max_value
- Storage usage: Two possible kinds of services:
- Monitoring one partition by service (metric are often named used and size)
- Monitoring multiple partitions by service and each metric corresponds to a partition name
In this two cases, the performance data returned by storage plugins have to correspond to this format:status information | metric_name=valueunit;warning_threshold;critical_threshold;min_value;max_value metric_name_2=value...
Traffic : Trafic standard reports use two metrics in parameters: one for the inbound trafic and one for the outbound trafic. Your plugins must return these two metrics, the name does not matter.:
*any status information* | **$inboundTrafic=$value$unit;$warning_threshold;$critical_threshold;$min_value;$max_value $outBoundTrafic=...**
It is important for the reporting that units of data collected by the monitoring solution are known, mastered and equals for same type of data, here are different units regarding to the data collected:
- Time: second
- Traffic: bits/sec
- Storage: Bytes
- Memory/Swap: Bytes
Centreon objects configuration’s best practices¶
In Centreon MBI, each report model has several parameters that allow the generation of customized documents according to your business requirements.
Different types of parameters can be found for each report model:
- A main object on which the report will be generated. It could be:
- A host
- A host group: functional group defined in Centreon to classify hosts by customer, application, business unit, country, ...
- Several host groups
- A timeperiod (or “Business hours” also called “Live service”) on which the statistics will be calculated.
- Filters to take into account only specific types of equipments, services and metrics from selected host groups:
- Host categories: allows to classify hosts in technical groups in order to determine the type or the technical function of a host; For example: Linux servers, Windows servers, Cisco routers, printers, ...
- Service categories: Allows to define the type of a service: CPU, physical memory, storage, ...
- Metrics: performance measures collected by the services (monitoring indicators). One monitoring service can collect several metrics, the metric names and units are not normalized (for example: a CPU type service can collect only one metric named “cpu_average” defined in percentage and another CPU type service can collect one metric by CPU core configured on the equipment), so it can be necessary, at the report generation, to select the metrics that must be taken in account in the statistic calculation.
Host groups and categories¶
The definitions of host groups and categories listed in the previous chapter are defined according to the best practices established by Centreon.
However, the groups and categories that you are going to create must correspond to your business needs.
If you need to report the number of alerts generated by IT field, with a detailed distribution by type of equipement, it could be necessary to define host groups and categories using this method:
- Host groups : Databases, Applications, Security, Network, Mail ...
- Host categories : DB2-Servers, MySQL-Servers, Oracle-Servers, SQL-Servers ...
Here is an exemple of statistics that you could get using these groups and categories :
The host group is the first analysis axis, the host category allows to analyse the statistics in subsets.
In the same way, we can analyse the statistics using the following dimensions:
- By country (host group) with a distribution of data by type of network equipment (host category)
- By country (host group) with a distribution of data by customer (host category)
- By customer (host group) with a distribution of data by country (host category)
- By customer (host group) with a distribution of data by application server (host category)
There is no standard rules to define host groups and categories. It must be adapted to your reporting needs.
How to create these relations ?
- The relation between hosts and host groups is made in the menu Configuration > Hosts > Host groups in the interface Centreon. To link host groups to a specific host, you can also use the Tab Relations in the host add/modification form.
- The relation between hosts and host categories is made in the menu Configuration > Hosts > Categories. To link host categories to a specific host, you can also use the Tab Relations in the host add/modification form.
The service categories allow to organize services (monitoring indicators) into subsets. The most common usage of the service categories is to define categories based on service types CPU, physical memory, storage, Process-Oracle, DNS, Process-Websphere, ...
This kind of configuration allows, for example, to:
- Compare the number of alerts generated by each type of service.
- Select the service category that indicates a storage usage information when you need to generate a capacity report.
Just like the host groups and categories, the service categories must be defined according to your reporting needs.
For example: If you need to analyse the storage space allocated and used by type of DBMS or type of application, you may need to create several service categories. Instead of using only one service category named “Storage” or “Disk” you could create these service categories :
- “Operating system”
- “SQL Server”
Here is an exemple of statistics that you could get using these service categories :
The relation between services and service categories is made in the menu Configuration > Services > Categories in the interface Centreon. To link service categories to a specific service, you can also use the Tab Service Extended Info the service add/modification form.
It is highly recommanded to manage service categories only using service templates.