time Time Drift - feedback

[T.Konopka 8]

Hello!

We need feedback about time synchronization and drifting.

Some of you confirmed that there was a slight improvement with 4.136, but this has not solved the issue completely.

How about 4.140 and 4.141?

Please, state the version of your system in your posts and how much the time is drifting or if it is not drifting at all.

Thanks!

[kunoch 27]

HCL, 4.140

I had a drift of about -8 minutes over two months in manual setting

Edited October 31, 2017 by kunoch

[Guest skarotech]

Mine has a drift of 2 minutes after a reboot 3 days ago...v4.140

 

as someone how has worked in IT for many many years, this is the whole idea around using NTP. While Fibaro has a NTP setting, why is this not checked and adjusted on a daily basis? Perhaps it is and there is another major issue causing the drift ?

 

 

Edited October 31, 2017 by skarotech

[jakub.jezek 370]

HC2 (testing unit 4.141 beta) -> + 1,5 minutes after restart 5 day ago. Using ntp server: ntp.nic.cz

HCL (testing unit 4.141 beta) -> - 6 minutes after restart 4 days ago. Using ntp server: ntp.nic.cz

 

HCL (showroom 4.140) -> -2 minutes after restart 5 days ago. Using ntp server: ntp.nic.cz

[Vinisz 15]

4.141 Beta , time download from internet setting, uptime: 1d, 20h, 45m

Time on the HC2 is 43 seconds ahead of internet time now. (ntp server setting: pool.ntp.org)

 

[alschout 0]

HC2 4.140

Manual timesetting since 5 weeks.

No drift at all.

Time is stable.

[northyorks 35]

4.140 - 5 mins slow even after reboot set on automatic time.

[FunkFib 14]

HC2 4.140

Timerdifference 4min. + after 4 weeks without syncronisation.

[japasu 15]

HC2 4.140

Uptime: 12d, 21h, 27m

Download time is ON

Time drift: + 2m, 40s

 

[cag014 730]

HC2 - 4.140

Up time  2d, 21h, 31m

Set date and time = manual settings

 

here is my debug information from the system:
[DEBUG] 21:59:27: Time Synchronization [200] [0] Drift = 59 Sec. [ 00:59 ]
[DEBUG] 21:59:16: Time Synchronization [200] [0] Drift = 71 Sec. [ 01:11 ]
[DEBUG] 21:58:58: Time Synchronization [200] [0] Drift = 83 Sec. [ 01:23 ]

looks like time drifting 12 seconds per day !

I've noticed that time drift depends on system load... have added scene that runs every 35 seconds and drift increased.

 

Edited November 2, 2017 by cag014

[Guest cuprum]

Feedback - initially i think, i asked it from You There was simple question: Is there some plans to implement *standard* solution - NTP*D* (daemon) based one - in Fibaro boxes?

PS. For me personally also "no" is also acceptable answer, with short explanation, like no room on disk.

[drboss 107]

HC2 4.140

14-20 sec by day, depends of system load

[hTiger1 50]

HCL, no time drift.

 

Please login or register to see this image.

/monthly_2017_11/capture20171103101927125.png.3c0b6ae995bac95458d684d2def72419.png" alt="capture20171103101927125.png.3c0b6ae995bac95458d684d2def72419.png" /> 

[Sankotronic 2,370]

My HC2 on 4.140 with NTP settings enabled after 5 weeks is now 10 minutes ahead of real time! It was rebooted two weeks ago by my wife because it stopped working (get frozen), but that didn't help to reduce this time drift difference.

 

Please login or register to see this attachment.

 

It seems like my HC2 is one of the kind with time drifting!  

 

Unfortunately I don't see that anything new is done with this two new betas regarding this problem. Shame!  

[samuel 27]

Which NTP server are being used, by each others setup? Would it change if the NTP Time server will be changed?

[10der 832]

Please login or register to see this image.

 

Server Status:30d, 22m

 

[Igy 23]

35 seconds behind after 1d and 21h from last restart on v 4.142 - HC2, the gap is increasing every day forcing me to restart device periodically. 

[Guest cuprum]

On 11/10/2017 at 12:16 PM, samuel said:

Would it change if the NTP Time server will be changed?

Unfortunately not

Actually, we do not know full information, but seems, that HC is syncing time only when it reboots. So, if Your box is near-to-real aft reboot the time server does not matter anymore.

Yep, there is little possibility, that configured server does not answers and the fast-drifting default time from RTC is used - but this situation is detectable easily with comparing times after boot. If the time is then quite realistic, then server-change clearly will not help. If differs - yep, change may help.

 

PS. most of time-servers are actually "pools" anyway, so even when multiple times attempt to sync from the same "timeserver by DNS name", it's really mean times from different physical clock devices

[Pielhaas 1]

3 minutes in 12 days after setting it manually on v 4.140 - HC2

[tinman 1,070]

typical drift of this intel board is 13min/year at 25° (2sec/day), plus 3sec/day for rtc crystal at 40°, gives 4-5sec/day.

Unfortunately i rebooted my test system 3day ago without writing down the collected values, but since then the drift is at 6sec/day (board / cmos bat 18months old), which is ok.