CIB: active_service_checks_(1min|5min) returns invalid values in large environments

[TheFireMike]

Describe the bug

In large environments the active_service_checks_(1min|5min) counter of the CIB returns invalid values.

Examples:

Result of REST API endpoint /status/CIB:

...
"active_service_checks_15min": 130324,
"active_service_checks_1min": 5,
"active_service_checks_5min": 18920,
...

Same request 10 seconds later:

...
"active_service_checks_15min": 130158,
"active_service_checks_1min": 4,
"active_service_checks_5min": 25,
...

Again 10 seconds later:

...
"active_service_checks_15min": 126361,
"active_service_checks_1min": 2,
"active_service_checks_5min": 1598,
...

To Reproduce

1. Create a large Icinga environment
2. Query the REST API Endpoint /status/CIB

Expected behavior

A response which returns how many checks were actually active during the last minute / last 5 minutes.

Your Environment

• Version used: v2.13.1

[cmaile]

Same behavior with version 2.13.4.

[carraroj]

ref/NC/815287

[Al2Klimov]

Icinga doesn't interpolate anything here as the statistics buffers are large enough. Are you sure you don't just have checks spikes like this?