On Tue, 02 May 2017 00:11:58 +0100, Mike Tomlinson wrote:
> En el artículo <xQLNA.1089033$nT1.2...@fx44.am4>, Johnny B Good
> <
johnny...@invalid.ntlworld.com> escribió:
>
>> Refitting the covers will mitigate the problem
>
> There /are/ no covers. It's a bare board on the bench with five hard
> discs attached :)
>
>
https://www.dropbox.com/s/hy928i09fiqbq8z/naked_pc.jpg?dl=0
>
> It's the one with a zfs array that I recently posted about in
> uk.comp.homebuilt.
>
>> you'll have to temporarily relocate the
>>radio clock to another room for a short while (half an hour or so?) to
>>allow it to sync back up
>
> oh, I can wait. I've got the PC clock which is NTP sync'd.
>
> I just hadn't connected the loss of radio clock sync with setting up the
> barebones PC.
Ah! It's so easy to overlook what in hindsight becomes the bleedin'
obvious. :-) BTDTAGTBTS. Mind you, I earned my T-shirt nearly a decade
ago when I realised that the radio controlled clock built into the I.T.
Works weather station was only syncing up overnight when I was able to
shut down the laptop I was using for late night/wee small hours PVR duty
(I very rarely left the desktop PC on overnight to complete any PVR tasks
- the laptop was a better use of the electricity for that purpose).
The Wx station is less than 2 foot away from the laptop (and ditto the
desktop monitor which *does* get switched off overnight regardless). The
desktop sits about 4 foot away which is still too close to allow reliable
reception by the Wx station imo.
Since the computer clocks are synced to ntp servers, I just learned to
accept the creeping discrepancy in the Wx station's displayed time. My
venerable Casio DB360 wristwatch has settled down to drifting by less
than a second over the past 4 months so I use it as a "sanity check" that
the PC clock is staying synced to an ntp server.
However, one oddity I've just experienced with the computer clock is the
inability to trigger a manual resync. I've just tried to resync to reduce
the 3 second discrepancy with my watch back to the leap second induced 1
second behind error and it just claims an inability to connect to any ntp
server and slows the clock by half a second each attempt. I'm pretty
certain it's always behaved like this, yet it must manage to
automatically sync up since day to day (and this machine is rarely
powered off) it's usually just showing the 1 second lag with respect to
my watch due to my not being arsed to adjust for that leap second just
over 4 months ago. I'm running Linux Mint 17.1 KDE 64 for anyone
wondering what might be behind this peculiarity.
I'll check it tomorrow and see whether that 3 second discrepancy has
returned or not before considering the problem any further. However,
curiosity got the better of me and after wasting precious minutes
googling ntp servers, I eventually hit upon using "what is the correct
time now" which gave me this link: <
https://time.is/GMT>" which is 3
seconds behind my watch strongly suggesting that my watch, after more
than 3 months of being ahead by a second has now suddenly gained an extra
two seconds during the last week or three.
If it's my watch suddenly gaining an extra couple of seconds, I expect
to see the computer clock time go adrift by 3 seconds WRT the watch when
I check it tomorrow (I've currently got the PC clock set to one second
behind the watch). The Wx station, btw, is 7 seconds slow compared to
time.is (making it 9 seconds behind the computer clock).
--
Johnny B Good