At 22:48 02-01-03, David Boyes wrote: >I'll assume you're not running NTP for the obvious reason that it >negates the timer patch benefit, but we've seen some relationship to >frequency of disk I/O and network I/O to how bad the timekeeping >becomes. I haven't been able to pin it down either.
I admit doing many things only half these days... but from what I have seen NTP is not that bad. The check rate is reduced when it is close to the proper time. So if indeed the skew is caused by high I/O rates then it should only cause some extra NTP overhead short after such activity. PS It cannot be as bad as the bug in ReiserFS on the 2.4.7 kernel that makes it hit 20 times per second. Rob
