Re: Diversity and offensive terminology in RFCs

[Dave Cridland <[e-mail]>]

On Thu, 20 Sep 2018 at 18:33, Evan Hunt <[e-mail]> wrote:

> On Thu, Sep 20, 2018 at 12:51:18PM +0100, Dave Cridland wrote:
> > Back when I was even more clueless than I am today, and actually ran DNS
> > servers, we used the terms "primary" and "secondary" as a matter of
> course.
> > Secondaries copied the data from primaries.
> >
> > So far, so good.
> >
> > Then we added a third nameserver, and of course that must be the
> tertiary,
> > used only when *both* the primary and secondary had failed.
>
> This suggests that at the time, you may have been under the impression
> that the secondary server would be used only when the primary had failed?
>
>
Yes, that is exactly what I thought at the time, based on the terms.


> For whatever it's worth, I think of it as concentric rings, with "primary"
> referring to the center ring and "secondary" to the next ring out. Data
> originates in the primary ring, and migrates to and through the secondary
> ring, and we name the servers "primary" or "secondary" depending on which
> ring they're in.
>
> I never thought of "master" as having political or racial implications
> (though, of course, as a white American guy, I do excel at overlooking
> such things).  I thought of it as comparable to "master key".  But at some
> point it crossed my mind that nobody ever talks about "slave keys", and if
> the original choice of terminology had been so innocuous, then why don't
> call them "master" and "copy"?  And then I started to see how these terms
> could indeed come across as unwelcoming to people of some backgrounds.
>
> And since "primary/secondary" works fine, I now try to use it, at least in
> professional writing.  In casual conversation, old habits take longer to
> break.
>

My point was that primary/secondary really didn't work for me.

I quite like Tony Finch's suggestion of "master/replica" though, for DNS in
particular.