Most hadoop ecosystem daemons use hostname.
One machine can have multiple ip's. And there is no way to tell what the
correct/authoritative ip is. Each machine has one and only one host name
(it can have more dns records pointing to it but only on hostname). Host
name's are supposed to be unique identifiers of the box/machine. So while
it adds a little work with your dns system to make sure that forward and
reverse dns' match up, it's the more unixy way of naming a machine.
On Tue, Jun 26, 2012 at 10:23 AM, Jean-Daniel Cryans <[EMAIL PROTECTED]>wrote:
> I don't recall specific discussions about it, it might just have been
> for readability of the user interfaces.
> On Tue, Jun 26, 2012 at 8:50 AM, Ben Lisbakken <[EMAIL PROTECTED]> wrote:
> > Hello --
> > I'm wondering if anyone knows why the hbase master broadcasts its
> > instead of its IP when it is telling the system where to connect to it?
> > Does anyone know if there is a specific reason behind this?
> > I'm asking because I am debating making a code change in hbase to force
> > master to report its IP and I am wondering if anyone is aware of a design
> > decision for why it isn't like this in the first place? (modifying
> > /etc/hosts and /etc/hostname is not a great solution for my particular
> > Thanks,
> > Ben