After 6 months of development new version of the monitoring system , whose code licensed under GPLv2. Zabbix consists of three basic components: a server for coordinating the execution of checks, generating check requests and collecting statistics; agents for performing checks on the side of external hosts; frontend for organizing system management.
To remove the load from the central server and form a distributed monitoring network, a series of proxy servers can be deployed, aggregating data on checking a group of hosts. Data can be stored in MySQL, PostgreSQL, TimescaleDB, DB2 and Oracle. Without agents, the Zabbix server can receive data via protocols such as SNMP, IPMI, JMX, SSH/Telnet, ODBC, and test the availability of Web applications and virtualization systems.
:
- A new type of agent is introduced - zabbix_agent2, written in the Go language and providing a plugin development framework for testing various services and applications. The new agent includes a built-in scheduler with support for flexible scheduling of checks and can track status between checks (for example, by keeping the connection to the DBMS open). To save traffic, sending received data in batch mode is supported. The new agent can be used to transparently replace the old one, so far only on the Linux platform;
- Added the ability to use and own handlers of actions and notifications when detecting failures of the checked services. Handlers can be written in JavaScript and used to call external notification delivery services or issue tracking systems. For example, you can write a handler to send problem messages to a corporate chat;
- Implemented official support for DBMS as a repository for audit data. Unlike previously supported
MySQL, PostgreSQL, Oracle and DB2, the TimescaleDB DBMS is specially optimized for storing and processing data in the form of a time series (slices of parameter values ββat specified time intervals, a record forms a time and a set of values ββcorresponding to this time). TimescaleDB allows you to significantly when working with such data, demonstrating an almost linear level of performance. In addition, TimescaleDB supports features such as automatic cleanup of old records; - specifications for the design of templates to standardize settings. The structure of the XML/JSON files has been changed to a form suitable for editing the template manually in a regular text editor. Existing templates are aligned with the proposed specifications;
- A knowledge base has been implemented to document the elements and triggers to be checked, which can be provided with a detailed description, an explanation of the goals for collecting information and instructions on how to act in case of problems;
- Enhanced capabilities for visualizing the state of the infrastructure are presented. Added the ability to change widget parameters with one click. Plot sets are optimized for display on wide screens and large wallboards. All widgets are adapted to display in the mode without titles. Added a new widget for displaying chart prototypes. A new aggregated view mode has been added to the widget with summary statistics of problems;
- Support for displaying data processed using various aggregate functions has been added to bar charts and graphs, which makes it more convenient to analyze data over long periods of time and simplifies planning. The following functions are supported: min,
max
avg
counting,
sum
first and
last; - Added the ability to automatically register new devices using PSK-keys (Pre-shared key) with encryption of settings for the added host;
- Added support for the extended JSONPath syntax, which allows organizing complex data preprocessing in JSON format, including aggregation and search operations;
- Added support for attaching descriptions to custom macros;
- The collection and definition of data related to WMI, JMX, and ODBC has been improved through the addition of new checks that return arrays of objects in JSON format. Also added storage support for VMWare and systemd services, as well as the ability to convert CSV data to JSON;
- The maximum limit on the number of dependent elements has been increased to 10 thousand;
- Support for new platforms has been added: SUSE Linux Enterprise Server 15, Debian 10, Raspbian 10, macOS and RHEL 8. An MSI agent package has been prepared for Windows. Added support for quick deployment of a monitoring system in an isolated container or in cloud environments AWS, Azure,
Google Cloud Platform,
Digital Ocean and Docker.
Source: opennet.ru