[Opendnssec-develop] Making ODS immune for localtime changes

[Jerry Lundström]

First off, very good work Rick!! Kudos!

On Nov 5, 2012, at 12:14 , Rick van Rein wrote:

>> This is worth looking at closely. One worry for me would be pushing
>> these changes out as we will then switch people from local time to
>> UTC which for most folks will be more than the one hour summer-time
>> switch.
> 
> Yes, I share that concern for one half of the globe.  We may have to
> play a Phileas Fogg on them; preparing the database with time zones
> on the safe side of UTC: "UTC-23h" or "UTC+23h" (not sure which),
> and so on.

I don't think we need "push" UTC on people, if we work by UTC in the database and the time calculations which its not affected by summer time then we could easily display the times in logs and command line output using the current timezone the system has configured.

>> Also note that InceptionOffset should not interfere with keys being
>> introduced; but maybe we should at least recommend that the publish
>> safety is at least one hour to cover DST (this is the supplied
>> default)?
> 
> Both variables would need our thinking through I suppose.  Setting them
> to 1h + the actually wished-for variance sounds to me like a workaround,
> but not an actual solution.  It's weird to have to set such variables
> to a high value just because there *may* be a time shift every now and
> then, and be delayed because of it throughout the rest of the year.

I agree with Rick here, it sounds like a workaround and I would really like to see a solution instead. We should not assume anything and we should be able to handle a system timezone change large then 1h.

/Jerry

--
Jerry Lundström - OpenDNSSEC Developer
http://www.opendnssec.org/