[arch-general] UTC

[Ralf Mardorf]

On Tue, 19 Feb 2013 01:40:10 +0100, Tom Gundersen <teg at jklm.no> wrote:
>> # timedatectl set-local-rtc 1
>> Failed to issue method call: Input/output error
>
> That message is disconcerting (at least the error message should have
> been clearer). Could you have a look in "journalctl" (as root) to see
> if you can get more information about what went wrong?
>
>> Sometimes I don't get an output when running "timedatectl set-local-rtc  
>> 1".

Resp. one time.

> That means it succeeded.

Hi Tom,

I don't know how to use journalctl, to get some useful information.

root at freebsd:/mnt/archlinux/home/rocketmouse # cat journalctl.txt
# journalctl | grep RTC
Feb 17 22:37:38 localhost kernel: RTC time: 21:37:32, date: 02/17/13
Feb 17 22:37:38 localhost kernel: rtc_cmos 00:04: RTC can wake from S4
Feb 18 00:01:52 localhost kernel: RTC time: 23:01:46, date: 02/17/13
Feb 18 00:01:52 localhost kernel: rtc_cmos 00:04: RTC can wake from S4
Feb 18 11:48:07 localhost kernel: RTC time: 10:48:01, date: 02/18/13
Feb 18 11:48:07 localhost kernel: rtc_cmos 00:04: RTC can wake from S4
Feb 18 12:44:33 localhost kernel: RTC time: 11:44:28, date: 02/18/13
Feb 18 12:44:33 localhost kernel: rtc_cmos 00:04: RTC can wake from S4
Feb 18 13:06:26 localhost kernel: RTC time: 12:06:20, date: 02/18/13
Feb 18 13:06:26 localhost kernel: rtc_cmos 00:04: RTC can wake from S4
Feb 18 13:58:44 localhost kernel: RTC time: 12:58:38, date: 02/18/13
Feb 18 13:58:44 localhost kernel: rtc_cmos 00:04: RTC can wake from S4
Feb 18 13:04:12 localhost systemd-timedated[291]: Failed to set RTC to  
local/UTC: Input/output error
Feb 18 14:38:10 localhost kernel: RTC time: 13:38:04, date: 02/18/13
Feb 18 14:38:10 localhost kernel: rtc_cmos 00:04: RTC can wake from S4
Feb 18 14:44:22 localhost kernel: RTC time: 13:44:16, date: 02/18/13
Feb 18 14:44:22 localhost kernel: rtc_cmos 00:04: RTC can wake from S4
Feb 18 19:17:11 localhost kernel: RTC time: 18:17:05, date: 02/18/13
Feb 18 19:17:11 localhost kernel: rtc_cmos 00:04: RTC can wake from S4
Feb 19 03:10:34 localhost kernel: RTC time:  2:10:27, date: 02/19/13
Feb 19 03:10:34 localhost kernel: rtc_cmos 00:04: RTC can wake from S4

# journalctl | grep UTC
Feb 17 22:37:38 localhost kernel: rtc_cmos 00:04: setting system clock to  
2013-02-17 21:37:33 UTC (1361137053)
Feb 18 00:01:52 localhost kernel: rtc_cmos 00:04: setting system clock to  
2013-02-17 23:01:47 UTC (1361142107)
Feb 18 11:48:07 localhost kernel: rtc_cmos 00:04: setting system clock to  
2013-02-18 10:48:02 UTC (1361184482)
Feb 18 12:44:33 localhost kernel: rtc_cmos 00:04: setting system clock to  
2013-02-18 11:44:28 UTC (1361187868)
Feb 18 13:06:26 localhost kernel: rtc_cmos 00:04: setting system clock to  
2013-02-18 12:06:21 UTC (1361189181)
Feb 18 13:58:44 localhost kernel: rtc_cmos 00:04: setting system clock to  
2013-02-18 12:58:39 UTC (1361192319)
Feb 18 13:04:12 localhost systemd-timedated[291]: Failed to set RTC to  
local/UTC: Input/output error
Feb 18 14:38:10 localhost kernel: rtc_cmos 00:04: setting system clock to  
2013-02-18 13:38:05 UTC (1361194685)
Feb 18 14:44:22 localhost kernel: rtc_cmos 00:04: setting system clock to  
2013-02-18 13:44:17 UTC (1361195057)
Feb 18 19:17:11 localhost kernel: rtc_cmos 00:04: setting system clock to  
2013-02-18 18:17:06 UTC (1361211426)
Feb 19 03:10:34 localhost kernel: rtc_cmos 00:04: setting system clock to  
2013-02-19 02:10:28 UTC (1361239828)

# journalctl
-- Logs begin at Sun 2013-02-17 22:37:38 CET, end at Tue 2013-02-19  
02:25:27 CET. --
Feb 17 22:37:38 localhost systemd-journal[128]: Allowing runtime journal  
files to grow to 185.2M.
[snip]

> If I remember correctly, one of the side effects of having your RTC in
> localtime is that nothing will write to the RTC automatically.

Correct for my new Arch Linux install.

> I know you don't want to hear about the problems with localtime, but
> to anyone else stumbling upon this: not having your RTC in UTC is
> broken and will cause problems.

It doesn't cause an issue in 10 years, but for sure using UTC usually is  
the better choice, if you don't want to have the local time for the BIOS,  
e.g. for timestamps when storing BIOS settings, that by the way are  
limited anyway.
For personal historical reasons currently all my *NIX expect local time.

Regards,
Ralf