[ntp:questions] Re: ntp sanity limit kills ntp daily

David Woolley david at djwhome.demon.co.uk
Fri May 20 19:56:46 UTC 2005


In article <cZqdnXESm5kSrxPfRVn-tA at comcast.com>,
Michael Ward <michaelward at sprintmail.com> wrote:

> clock looses some 1200 seconds while the system is busy making backups. 
>   When ntp figures this out, it refuses to update the clock because this 
> is classified as insane, and terminates.

You are losing clock interrupts.  You need to fix the underlying cause.
That will probably be your disk driver, however a co-factor on recent
Linux systems is the use of HZ=1000 rather than the normal HZ=100
(clock interrupts per second), on faster machines.  You should rebuild
the kernel with the more sensible interrupt rate.

Linux and Windows both have a poor reputation for losing clock interrupts.



More information about the questions mailing list