On Mon, Aug 17, 2015 at 11:20 AM, Miroslav Lichvar <mlich...@redhat.com> wrote: > It looks like the output was captured soon after chronyd was started. > With only 4 measurements spread over 9 seconds such large frequency > error wouldn't be unusual. What do you see when chronyd is running for > an hour? > > To further debug this it might be useful to configure chronyd to not > touch the clock by adding the noselect option to all servers specified > in the config file and enable the measurements and statistics logs > (see the logdir and log directives). > > It could be a bug in the clock source that is used on the BB. You > could try a different clock source if there is more than one > available.
I checked and I have a single clock source: debian@bbw1:~$ cat /sys/devices/system/clocksource/clocksource0/available_clocksource timer1 Note that I tried to start chrony in a clean state, without a chrony.drift file: debian@bbw1:~$ sudo systemctl stop chrony debian@bbw1:~$ sudo rm /var/lib/chrony/* debian@bbw1:~$ sudo reboot After over 30 minutes the machine that have is not adjusting the clock (servers with noselect), don't have any unreasonable estimate, but the one adjusting the clock still have very large errors: debian@bbw1:~$ uptime 00:39:28 up 37 min, 1 user, load average: 0.11, 0.04, 0.03 debian@bbw1:~$ chronyc sourcestats 210 Number of sources = 3 Name/IP Address NP NR Span Frequency Freq Skew Offset Std Dev ============================================================================== ns.rnl.tecnico.ulisboa.pt 4 3 715 -71491.055 17976.291 -16.6s 199ms adsl-62-48-148-210.ptprim 4 3 715 -71394.406 14362.602 -16.4s 230ms a88-157-128-22.cpe.netcab 4 3 956 -72689.414 28005.160 -17.2s 382ms debian@bbw1:~$ chronyc sources 210 Number of sources = 3 MS Name/IP address Stratum Poll Reach LastRx Last sample =============================================================================== ^* ns.rnl.tecnico.ulisboa.pt 3 8 377 233 +220ms[ -16.6s] +/- 11ms ^- adsl-62-48-148-210.ptprim 3 8 377 233 +218ms[ +218ms] +/- 83ms ^- a88-157-128-22.cpe.netcab 3 8 375 227 -169ms[ -169ms] +/- 69ms debian@bbw1:~$ chronyc tracking Reference ID : 193.136.164.1 (ns.rnl.tecnico.ulisboa.pt) Stratum : 4 Ref time (UTC) : Mon Aug 17 19:53:42 2015 System time : 493154176.000000000 seconds slow of NTP time Last offset : -16.801349640 seconds RMS offset : 13.358787537 seconds Frequency : 0.000 ppm fast Residual freq : -71491.055 ppm Skew : 1000000.000 ppm Root delay : 0.021884 seconds Root dispersion : 270.499451 seconds Update interval : 238.8 seconds Leap status : Normal Note this machine have the clock so long in the past because the default debian config does not enable steps, and the power on date is Jan 1 2000. This doesn't make any difference. Would you like remote access to the machine? I have a clean install of debian, so you can hack it at will. If so please send me your public key. Thanks, Nuno -- To unsubscribe email chrony-users-requ...@chrony.tuxfamily.org with "unsubscribe" in the subject. For help email chrony-users-requ...@chrony.tuxfamily.org with "help" in the subject. Trouble? Email listmas...@chrony.tuxfamily.org.
