Werk #19278: Fix esx_vsphere checks handling unavailable VMware metrics incorrectly

Component	Checks & agents
Title	Fix esx_vsphere checks handling unavailable VMware metrics incorrectly
Date	Mar 3, 2026
Level	Trivial Change
Class	Bug Fix
Compatibility	Compatible - no manual interaction needed
Checkmk versions & editions	2.6.0b1 Not yet released Checkmk Raw (CRE), Checkmk Enterprise (CEE), Checkmk Cloud (CCE), Checkmk MSP (CME) 2.5.0b1 Not yet released Checkmk Raw (CRE), Checkmk Enterprise (CEE), Checkmk Cloud (CCE), Checkmk MSP (CME) 2.4.0p23 Not yet released Checkmk Raw (CRE), Checkmk Enterprise (CEE), Checkmk Cloud (CCE), Checkmk MSP (CME)

VMware vSphere reports -1 for performance counter metrics that could not be collected in a given 20-second sample interval. Several ESX checks did not handle this sentinel value and passed it through as a real measurement:

• Disk IO SUMMARY: -1 values in disk latency metrics caused a crash with ValueError: Cannot render negative timespan
• Uptime: a -1 sample as the most-recent value produced a negative uptime
• VMKernel Swap: services showed -1 B for swap in, out, and used metrics
• Ramdisk [name]: a -1 sample caused an incorrect used-space value to be reported
• GPU Utilization [id]: utilization was reported as -1% instead of going UNKN
• Interface [name]: -1 network counter samples were included in the averaged rate

Invalid -1 samples are now excluded from all affected checks. Services report UNKN or omit the metric entirely when no valid sample is available in a check interval.

To the list of all Werks