Re: Redundancy - why not do it on Unix?

From: Martin Hamilton <martin@dont-contact.us>
Date: Fri, 07 Nov 1997 11:02:00 +0000

-----BEGIN PGP SIGNED MESSAGE-----

David Luyer writes:

| * setup a machine with an IP alias of your main cache machine somehow
| modified so it is slow at responding to ARP requests, or make a job
| insert the IP alias only when the other cache is detected as being down.

If your machines have two or more NICs you could have a system whereby
an interface per machine was used to check for up-ness of the other
machines and ARP used to step in temporarily during an outage. I
understand this is something the Linux high availability project is
working on (is this still vapourware? :-) but I suspect there's no
reason why it couldn't be done with most Unixes.

Having said that, I think there are a few pathological cases, e.g.
SunOS 4.1.x boxes tend to go a bit funny if they think someone else
is using their IP address ?

Cheerio,

Martin

-----BEGIN PGP SIGNATURE-----
Version: 2.6.3i
Charset: noconv

iQCVAwUBNGL1JdZdpXZXTSjhAQG7XgP/SYCxY9XUDgJfHLl7QGsrTuQzAs+NyXIa
E7yLa0eHyT89JVWgMuLvK98BgZ/5UiRyVZibTTgjsUDvcn2s7Ny3wNIw7B95zJyo
57ebftKU9G2KYP/xtJFoTw88a0EXP3W+MJSuAMI7J61o/XTTI8pCBDmQ6Z1zD4HP
xA8C2Y1QXfI=
=VBYK
-----END PGP SIGNATURE-----
Received on Fri Nov 07 1997 - 03:05:42 MST

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:37:28 MST