Roughtime servers list?

[Jeroen Massar]

Hi folks,

Is there already a roughtime servers list, so that we can add instances
there? :)

Greets,
Jeroen

[Adam Langley]

On Tue, Sep 20, 2016 at 11:41 AM, Jeroen Massar <jer...@massar.ch> wrote:

Hi folks,

Is there already a roughtime servers list, so that we can add instances
there? :)

There is a roughtime-servers.json checked into git that contains the Google server currently. There is not yet a process for adding more servers, although there are some ideas sketched at https://roughtime.googlesource.com/roughtime/+/HEAD/ECOSYSTEM.md#Curating-server-lists

Cheers

AGL

[Hanno Böck]

Hi,

On Tue, 20 Sep 2016 11:44:41 -0700
Adam Langley <a...@chromium.org> wrote:

> There is a roughtime-servers.json checked into git that contains the
> Google server currently. There is not yet a process for adding more
> servers, although there are some ideas sketched at
> https://roughtime.googlesource.com/roughtime/+/HEAD/ECOSYSTEM.md#Curating-server-lists

Not sure if this is answered anywhere and I missed it, but what I was
asking myself: If I would operate a server where would I get my time
from? From other servers? Is there a hierarchy of servers then? (gets
time just from some other roughtime server vs. gets time from an
"authoritative" source.) From NTP over insecure channels?
Or are only people supposed to run servers that somehow are connected
to atomic clocks or whatever it is that gives us the root of our time?

--
Hanno Böck
https://hboeck.de/

mail/jabber: ha...@hboeck.de
GPG: FE73757FA60E4E21B937579FA5880072BBB51E42

[Adam Langley]

On Tue, Sep 20, 2016 at 11:56 AM, Hanno Böck <ha...@hboeck.de> wrote:

Not sure if this is answered anywhere and I missed it, but what I was
asking myself: If I would operate a server where would I get my time
from? From other servers? Is there a hierarchy of servers then? (gets
time just from some other roughtime server vs. gets time from an
"authoritative" source.) From NTP over insecure channels?
Or are only people supposed to run servers that somehow are connected
to atomic clocks or whatever it is that gives us the root of our time?

We would not want to be in a position where hijacking pool.ntp.org lets you control a large number of Roughtime servers, so ideally they would have an independent time source (e.g. GPS or atomic clocks). There is also the question of correctly handling leap seconds. Roughtime specifies a 24-hour linear-smear but most NTP sources won't do that. (Granted that a second here or there is also, currently, more precision than we're aiming at.)

Cheers

AGL 

[Watson Ladd]

Why not bite the bullet for GPS native time, and have the client do
what it wants?

>
>
> Cheers
>
> AGL
>
> --
> You received this message because you are subscribed to the Google Groups
> "proto-roughtime" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to proto-roughti...@chromium.org.
> To view this discussion on the web visit
> https://groups.google.com/a/chromium.org/d/msgid/proto-roughtime/CAL9PXLw3tRH0AVBGuYB9n9vwar8ncaid-eGfy5CCj0%3DQgAugLw%40mail.gmail.com.



--
"Man is born free, but everywhere he is in chains".
--Rousseau.

[Adam Langley]

On Tue, Sep 20, 2016 at 12:13 PM, Watson Ladd <watso...@gmail.com> wrote:

Why not bite the bullet for GPS native time, and have the client do
what it wants?

There is certainly scope for adding some more information to the Roughtime response to aid clients (like a leap-second indication if a client is trying to sync a PLL). But UTC seems to be what people want to end up with, and it avoids clients having to ship with a leap-second table. (Clients that want to do something more fancy can ship with such a table and get GPS-time from Roughtime.)

Cheers

AGL