Microsoft describes an indeterminate Connection Monitor test as missing data in the Log Analytics workspace and directs operators to inspect metrics. It also documents a scale-set case: a specifically selected instance may be removed during the monitoring cycle without immediate recognition, leaving the monitor indeterminate. Microsoft recommends random instance selection within coverage levels to reduce this particular discovery problem. Microsoft Learn.
\nUse this review when an existing test becomes indeterminate, especially around a scale-in or deallocation. Identify the exact source instance, test group, configured coverage and observation window before changing network rules.
\nDSE recommends keeping a missing-data investigation separate from a confirmed failed-connectivity incident. Compare source-instance existence, monitoring coverage and workspace observations first. Where suitable, have the monitoring owner assess randomized coverage against the application’s monitoring requirements. Preserve the original selection and event timeline before adjusting the test.
\nCompare the selected instance’s lifecycle with metric and log timestamps. From an approved surviving source, run a harmless connectivity check to the same destination. Record that result separately from monitor recovery. Close the investigation only after the expected sources produce fresh observations and any remaining connectivity failure has its own evidence.
\n