selection of .GPS.

[Paul Theodoropoulos]

Something I'm still unclear on - my ntpsec server is happily rolling 
along with kernel PPS. Dandy. The GPS signal (fudged as close as I can 
get it) offset and jitter vacillate wildly, which from experience, is 
pretty much 'normal'.

However, as I watch ntpq -p, the selection of GPS vacillates wildly as 
well - at one moment selected (+), the next moment 'bad' (x), the next 
merely deselected (-).

e.g. (other peers elided simply for brevity):

root at RaspiPi3B+ NTPsec: ~ # ntpq -p
      remote                                   refid      st t when poll 
reach   delay   offset   jitter
=======================================================================================================
*SHM(2)                                  .PPS.            0 l    -    1  
377   0.0000   0.0000   0.0004
+SHM(0)                                  .GPS.            0 l    1    2  
377   0.0000  24.2791   8.5128


root at RaspiPi3B+ NTPsec: ~ # ntpq -p
      remote                                   refid      st t when poll 
reach   delay   offset   jitter
=======================================================================================================
*SHM(2)                                  .PPS.            0 l    1    1  
377   0.0000  -0.0004   0.0007
xSHM(0)                                  .GPS.            0 l    -    2  
377   0.0000  32.2590  10.7885


root at RaspiPi3B+ NTPsec: ~ # ntpq -p
      remote                                   refid      st t when poll 
reach   delay   offset   jitter
=======================================================================================================
*SHM(2)                                  .PPS.            0 l    1    1  
377   0.0000  -0.0001   0.0006
xSHM(0)                                  .GPS.            0 l    -    2  
377   0.0000  32.1088   9.0393

root at RaspiPi3B+ NTPsec: ~ # ntpq -p
      remote                                   refid      st t when poll 
reach   delay   offset   jitter
=======================================================================================================
*SHM(2)                                  .PPS.            0 l    -    1  
377   0.0000   0.0011   0.0013
-SHM(0)                                  .GPS.            0 l    1    2  
377   0.0000   8.2362  22.5395

root at RaspiPi3B+ NTPsec: ~ # ntpq -p
      remote                                   refid      st t when poll 
reach   delay   offset   jitter
=======================================================================================================
*SHM(2)                                  .PPS.            0 l    1    1  
377   0.0000  -0.0051   0.0051
xSHM(0)                                  .GPS.            0 l    -    2  
377   0.0000  16.0547  17.6070

My question is, is this rapid selection/deselection 'harmful' to my 
overall accuracy? Would it be better to just set 'noselect' on GPS, to 
avoid the constant use/disuse of it? Or is worth listening to, even with 
other stable offsite peers?

-- 
Paul Theodoropoulos
www.anastrophe.com