When the clocks went back, looking in the log for entries at 01:49:02 brought a very great surprise: in one Mac, nearly 4 million entries in a single second.
time
Apparently based on Mach absolute time, log entry times are converted to wallclock times. This exposes them to the vagaries of time zones, seasonal adjustments, and periodic wallclock adjustments. Here’s how all that works, and can confuse.
How to combine the time of interest with waypoints to reduce 100,000 log entries to just a handful, and discover what you’re looking for in the log.
Early Macs normally had their clocks synced manually, until System 8.5 introduced support for NTP. That later switched to a proprietary service, timed, in macOS 10.13 High Sierra.
This compares short time intervals obtained from log entry timestamps obtained from the log show command via Ulbow, those from LogUI using OSLog, and Mach Absolute Time.
Getting the date and time stamp of log entries to use rounded microseconds, and how to ensure a log extract uses the current time zone throughout.
It turns out that ‘nanosecond’ times introduced in LogUI are largely artefact. Is higher resolution timing really needed, and how can it be obtained?
How to ensure your Mac’s clock and time is as accurate as possible, what can cause problems, and how it works using the timed daemon, as revealed in the log.
Although Macs can resolve time to nanoseconds, this isn’t apparent from tests writing log entries very rapidly. This explains what is probably happening.
Although there are important differences between Intel and Apple silicon Macs, both can resolve time to nanoseconds. So can this new version of LogUI.
