Incorrect Nagios PerfData Parsing in check_icmp Probes

Hello Observium Team,

I would like to report a bug in the Nagios performance data parser used by the Probes functionality in Observium Professional.

Problem description

When configuring an ICMP probe using the check_icmp plugin, Observium incorrectly parses some of the performance metrics returned by the plugin.

For example, the plugin produces the following valid Nagios PerfData output:

rta=1.522ms;100.000;200.000;0; pl=0%;5;20;; rtmax=2.812ms;;;; rtmin=1.345ms;;;;

Instead of identifying the four individual metrics (rta, pl, rtmax, and rtmin), Observium incorrectly creates a metric named:

;5;20;; rtmax

As a result, packet loss (pl) and minimum RTT (rtmin) are not properly detected or graphed.

Root cause

I traced the issue to the parse_nagios_perfdata() function in:

includes/entities/probe.inc.php

The existing regular expression allows whitespace and portions of subsequent metrics to be consumed as part of the preceding metric's optional fields.

Proposed fix

I modified the parser to correctly separate metrics at whitespace boundaries while preserving support for quoted metric labels, optional threshold fields, and GAUGE/COUNTER types.

After applying the patch locally, Observium successfully detects and displays all four metrics independently:

• rta – Round Trip Average

• pl – Packet Loss

• rtmax – Round Trip Maximum

• rtmin – Round Trip Minimum

I have verified that all four metrics are being stored in RRD files and displayed correctly in the web interface over multiple polling cycles.

Attached: observium-probe-perfdata-fix.patch

The patch contains the changes to parse_nagios_perfdata() relative to SVN revision 15424.

The fix has been successfully tested with check_icmp, although I have not tested it against every possible Nagios PerfData format.

Could you please review the attached patch and consider incorporating a fix into an upcoming Observium release?

Thank you.