There is a new field present in slow query log lines and profiler entries called "cpuNanos". This represents the total CPU time spent by the operation's thread in nanoseconds. This field is only included on Linux systems and is absent otherwise.

This field is independent of the cpuNanos field that is already present in the "operationMetrics" slow query/profiler section enabled by Serverless resource consumption metrics. The two values may differ slightly based on how they are collected, but not significantly.

Description of Linked Ticket

We collect per-operation CPU time (SERVER-51667) but we do not enable this by default. Collect this by default so that it can be reported in slow query logs and the profiler.