Best practices question

Paul Theodoropoulos paul at anastrophe.com
Wed Jun 26 06:12:16 UTC 2019


On 6/25/2019 20:06 PM, Patrick wrote:
> An ntpsec direct to the RasPi Hat configured with:
> 	refclock nmea        refid GPS path /dev/ttyAMA0 baud 9600 minpoll 4 maxpoll 4 ppspath /dev/pps0 flag1 1 mode 1 time2 0.480
>
> Yields the following...
>
>       remote           refid      st t when poll reach   delay   offset   jitter
> ===============================================================================
> oNMEA(0)         .GPS.            0 l    7   16  377   0.0000  -0.0026   0.0002
> +17.253.82.253   .SHM.            1 u   31   32  377   2.5711  -0.0826   0.0126
> +118.189.138.5   .NMC1.           1 u   31   32  377   4.2364  -1.5600   2.3600
> +218.186.3.36    .GPS.            1 u   23   32  377   2.5162  -0.1025   0.0106
Interesting, thanks Patrick. I had to do a lot of futzing around to get it 
working, this was the only combination that got it going:

refclock nmea refid GPS path /dev/ttyAMA0 baud 115200 minpoll 4 maxpoll 4 
ppspath /dev/pps0 flag3 0 flag1 1 time1 0.0

But unfortunately, my stability went to hell, even after letting it settle 
in for a few hours. Entirely likely that my 'this was the only thing that 
worked' translates into 'I'm wildly off the mark in properly setting it 
up', but I seem to be out of my depth in that regard.

The one truly odd side-effect however is that ntpd is performing 'disk' 
I/O intermittently (confirmed via iotop), which is showing up in sar output -
22:15:05        CPU     %user     %nice   %system   %iowait %steal     %idle
22:20:05        all      0.02      0.00      0.02 0.03      0.00     99.93
22:25:05        all      0.01      0.00      0.01 0.04      0.00     99.94
22:30:53        all      0.02      0.00      0.02 0.05      0.00     99.91

that in turn is increasing the enclosure temp, and throwing things further 
out of kilter. My existing stable units never show any iowait. The 'why' 
of it I've been unable to track down. But for now I'm going back to the 
'old' setup...It was an interesting experiment though!

-- 
Paul Theodoropoulos
www.anastrophe.com



More information about the users mailing list