12. Regular expressions

12. Regular expressions

Overview

Perl Compatible Regular Expressions (PCRE, PCRE2) are supported in Zabbix.

There are two ways of using regular expressions in Zabbix:

manually entering a regular expression
using a global regular expression created in Zabbix

Regular expressions

You may manually enter a regular expression in supported places. Note that the expression may not start with @ because that symbol is used in Zabbix for referencing global regular expressions.

It’s possible to run out of stack when using regular expressions. See the pcrestack man page for more information.

Note that in multi-line matching, the ^ and $ anchors match at the beginning/end of each line respectively, instead of the beginning/end of the entire string.

Global regular expressions

There is an advanced editor for creating and testing complex regular expressions in Zabbix frontend.

Once a regular expression has been created this way, it can be used in several places in the frontend by referring to its name, prefixed with @, for example, @mycustomregexp.

To create a global regular expression:

Go to: Administration → General
Select Regular expressions from the dropdown
Click on New regular expression

The Expressions tab allows to set the regular expression name and add subexpressions.

All mandatory input fields are marked with a red asterisk.

Parameter		Description
Name		Set the regular expression name. Any Unicode characters are allowed.
Expressions		Click on Add in the Expressions block to add a new subexpression.
	Expression type	Select expression type: Character string included - match the substring Any character string included - match any substring from a delimited list. The delimited list includes a comma (,), a dot (.) or a forward slash (/). Character string not included - match any string except the substring Result is TRUE - match the regular expression Result is FALSE - do not match the regular expression
	Expression	Enter substring/regular expression.
Delimiter		A comma (,), a dot (.) or a forward slash (/) to separate text strings in a regular expression. This parameter is active only when “Any character string included“ expression type is selected.
Case sensitive		A checkbox to specify whether a regular expression is sensitive to capitalization of letters.

A forward slash (/) in the expression is treated literally, rather than a delimiter. This way it is possible to save expressions containing a slash, without errors.

A custom regular expression name in Zabbix may contain commas, spaces, etc. In those cases where that may lead to misinterpretation when referencing (for example, a comma in the parameter of an item key) the whole reference may be put in quotes like this: “@My custom regexp for purpose1, purpose2”.
Regular expression names must not be quoted in other locations (for example, in LLD rule properties).

In the Test tab the regular expression and its subexpressions can be tested by providing a test string.

Results show the status of each subexpression and total custom expression status.

Total custom expression status is defined as Combined result. If several sub expressions are defined Zabbix uses AND logical operator to calculate Combined result. It means that if at least one Result is False Combined result has also False status.

Default global regular expressions

Zabbix comes with several global regular expression in its default dataset.

Name	Expression	Matches
File systems for discovery	`^(btrfs\|ext2\|ext3\|ext4\|jfs\|reiser\|xfs\|ffs\|ufs\|jfs\|jfs2\|vxfs\|hfs\|refs\|apfs\|ntfs\|fat32\|zfs)$`	“btrfs” or “ext2” or “ext3” or “ext4” or “jfs” or “reiser” or “xfs” or “ffs” or “ufs” or “jfs” or “jfs2” or “vxfs” or “hfs” or “refs” or “apfs” or “ntfs” or “fat32” or “zfs”
Network interfaces for discovery	`^Software Loopback Interface`	Strings starting with “Software Loopback Interface”.
	`^lo$`	“lo”
	`^(In)?[Ll]oop[Bb]ack[0-9._]$`	Strings that optionally start with “In”, then have “L” or “l”, then “oop”, then “B” or “b”, then “ack”, which can be optionally followed by any number of digits, dots or underscores.
	`^NULL[0-9.]$`	Strings starting with “NULL” optionally followed by any number of digits or dots.
	`^[Ll]o[0-9.]$`	Strings starting with “Lo” or “lo” and optionally followed by any number of digits or dots.
	`^[Ss]ystem$`	“System” or “system”
	`^Nu[0-9.]$`	Strings starting with “Nu” optionally followed by any number of digits or dots.
Storage devices for SNMP discovery	`^(Physical memory\|Virtual memory\|Memory buffers\|Cached memory\|Swap space)$`	“Physical memory” or “Virtual memory” or “Memory buffers” or “Cached memory” or “Swap space”
Windows service names for discovery	`^(MMCSS\|gupdate\|SysmonLog\|clr_optimization_v2.0.50727_32\|clr_optimization_v4.0.30319_32)$`	“MMCSS” or “gupdate” or “SysmonLog” or strings like “clr_optimization_v2.0.50727_32” and “clr_optimization_v4.0.30319_32” where instead of dots you can put any character except newline.
Windows service startup states for discovery	`^(automatic\|automatic delayed)$`	“automatic” or “automatic delayed”

Examples

Example 1

Use of the following expression in low-level discovery to discover databases except a database with a specific name:

^TESTDATABASE$

Chosen Expression type: “Result is FALSE”. Doesn’t match name, containing string “TESTDATABASE“.

Example with an inline regex modifier

Use of the following regular expression including an inline modifier (?i) to match the characters “error”:

(?i)error

Chosen Expression type: “Result is TRUE”. Characters “error” are matched.

Another example with an inline regex modifier

Use of the following regular expression including multiple inline modifiers to match the characters after a specific line:

(?<=match (?i)everything(?-i) after this line\n)(?sx).*# we add s modifier to allow . match newline characters

Chosen Expression type: “Result is TRUE”. Characters after a specific line are matched.

g modifier can’t be specified in line. The list of available modifiers can be found in pcresyntax man page. For more information about PCRE syntax please refer to PCRE HTML documentation.

Regular expression support by location

Location		Regular expression	Global regular expression	Comments
Agent items
	eventlog[]	Yes	Yes	`regexp`, `severity`, `source`, `eventid` parameters
	log[]			`regexp` parameter
	log.count[]			`regexp` parameter
	logrt[]		Yes/No	`regexp` parameter supports both, `file_regexp` parameter supports non-global expressions only
	logrt.count[]		Yes/No
	proc.cpu.util[]		No	`cmdline` parameter
	proc.mem[]
	proc.num[]
	sensor[]			`device` and `sensor` parameters on Linux 2.4
	system.hw.macaddr[]			`interface` parameter
	system.sw.packages[]			`package` parameter
	vfs.dir.count[]			`regex_incl`, `regex_excl`, `regex_excl_dir` parameters
	vfs.dir.size[]			`regex_incl`, `regex_excl`, `regex_excl_dir` parameters
	vfs.file.regexp[]			`regexp` parameter
	vfs.file.regmatch[]
	web.page.regexp[]
SNMP traps
	snmptrap[]	Yes	Yes	`regexp` parameter
Item value preprocessing		Yes	No	`pattern` parameter
Functions for triggers/calculated items
	count()	Yes	Yes	`pattern` parameter if `operator` parameter is regexp or iregexp
	countunique()	Yes	Yes
	find()	Yes	Yes
	logeventid()	Yes	Yes	`pattern` parameter
	logsource()	Yes	Yes	`pattern` parameter
Low-level discovery
	Filters	Yes	Yes	Regular expression field
	Overrides	Yes	No	In matches, does not match options for Operation conditions
Action conditions		Yes	No	In matches, does not match options for Host name and Host metadata autoregistration conditions
Web monitoring		Yes	No	Variables with a regex: prefix Required string field
User macro context		Yes	No	In macro context with a regex: prefix
Macro functions
	regsub()	Yes	No	`pattern` parameter
	iregsub()	Yes	No	`pattern` parameter
Icon mapping		Yes	Yes	Expression field
Value mapping		Yes	No	Value field if mapping type is `regexp`