[ntp:questions] Regular spike_detect on syslog

Jan Ceuleers jan.ceuleers at computer.org
Sun Sep 16 14:34:24 UTC 2018

On 16/09/18 14:15, Sean Austin Critica wrote:
> Can I directly observe these sources and see which ones are stable
> (maybe by dumping them periodically, remotely from a machine with a
> known stable clock)? The OS in this case is RedHat EL 7.

"Directly observe": no. But you can look at the kernel log to see which
clock source the OS has selected:

# dmesg | grep clocksource
[    0.000000] clocksource: refined-jiffies: mask: 0xffffffff
max_cycles: 0xffffffff, max_idle_ns: 7645519600211568 ns
[    0.000000] clocksource: hpet: mask: 0xffffffff max_cycles:
0xffffffff, max_idle_ns: 79635855245 ns
[    0.053021] clocksource: jiffies: mask: 0xffffffff max_cycles:
0xffffffff, max_idle_ns: 7645041785100000 ns
[    0.181027] clocksource: Switched to clocksource hpet
[    0.198695] clocksource: acpi_pm: mask: 0xffffff max_cycles:
0xffffff, max_idle_ns: 2085701024 ns
[    0.906180] clocksource: tsc: mask: 0xffffffffffffffff max_cycles:
0x2879c5f06f2, max_idle_ns: 440795220049 ns
[    1.931401] clocksource: Switched to clocksource tsc

You can then look in your CPU's manual to determine whether the selected
clocksource is influenced by SpeedStep (or whatever dynamic CPU speed
changes are called these days).

HTH, Jan

More information about the questions mailing list