[ Date Index ] [ Thread Index ] [ <= Previous by date / thread ] [ Next by date / thread => ]
On Sat, 10 Jan 2015, Simon Waters wrote:
On 2015-01-10 01:51, bad apple wrote:http://www.wired.com/2015/01/torvalds_leapsecond/Did anyone actually have any servers affected by last leap second? I saw the fuss, but nothing I was responsible for glitched.With Gordon on the GMT/BST thing couple of Debian boxes did stupid CRON things on various transitions, on inspection one of them had coded their CRON scripts to try and cope with BST transition, when Vixie CRON already did that correctly.Bug report was basically "cut the messing about and it will work better".Really though there is no reason to run Linux servers in anything by GMT, and just set environment variable or settings for program output to have the right timezone. May require re-education of users. Bit of a pain when the displayed time is an hour out, but saves a whole lot of messing once you are trying to track failures across multiple timezones."What is this log entry in?" "GMT like all the others".Although I confess getting it all to GMT is on the to do list for some of my current servers. For one off servers getting the time zone right is probably easier than using GMT. YMMV
s/GMT/UTC/g But it's all relative.Time just does my head in. Even UTC is a made-up acronym and the boffins keep on re-defining the idea of a second.
UTC - Universal Coordinated Time.But in France, it's really TUC - Temps Universel Coordonné, or in English more properly CUT - Coordinated Universal Time, so UTC is the compromose.
However the earth keeps on turning, although it's slowing down, so we need leap seconds to bring UTC in-step with UT1 (mean solar time- although the sun isn't used to actually calculate it anymore) else our sundials would drift.
GMT is historically closer to UT1. UT1 can be as much as 0.9 seconds different from UTC.
Gordon
-- The Mailing List for the Devon & Cornwall LUG http://mailman.dclug.org.uk/listinfo/list FAQ: http://www.dcglug.org.uk/listfaq