On Fri, Nov 25, 2016 at 05:11:08PM -0800, Denny Page wrote: > I believe the stddev issue with multiple interfaces may have been leftover > sampling from prior runs. I cleared all the point data, and things looked > better. Sorry about that. > > The priority issue remains, but I don’t think it’s a big deal per se. When > running with scheduling priority (-P 9), I see about 30% of ‘D H’ for the > fast units. Running without priority I see virtually none (all ‘H H’). I > don’t believe that this is much of an issue, because scheduling priority is > really only needed when daemon timestamps are in use. With hardware or driver > timestamps, it isn't necessary.
I'm glad you found a workaround, but I'd still like to understand what exactly is happening and whether waiting for the TX timestamp would fix it. We might want to add an option for that. Can you please try the patch from the following post and see if it turns the 'D H' measurements into 'H H' when running with -P 9? https://www.mail-archive.com/chrony-dev@chrony.tuxfamily.org/msg01453.html -- Miroslav Lichvar -- To unsubscribe email chrony-dev-requ...@chrony.tuxfamily.org with "unsubscribe" in the subject. For help email chrony-dev-requ...@chrony.tuxfamily.org with "help" in the subject. Trouble? Email listmas...@chrony.tuxfamily.org.
