[squid-users] Question on sibling relationships

Hi,

        I'd like to clarify how squid accesses the Internet when its part of a hierarchy. I've read the Squid config guides and FAQ and it doesn't directly address the questions. If a proxy is part of a hierarchy, say it has some siblings and a parent, how does that particular proxy handle a network error when, say, its link to its parent dies, but the link toward its siblings (and assuming its siblings connection to the internet is ok) is ok? Does the proxy then proxy requests through its siblings? I know that the proxy queries its siblings before its parent, and it only queries the parent if its siblings do not return the object requested.

        This is the scenario I'm facing. The proxy cluster is configured as follows:

 ISP1 ISP2 ISP3
  | | |
  | | |
+-+-+ +-+-+ +-+-+
| A |--sibling--| B |--sibling--| C |
+-+-+ +---+ +-+-+
  | |
  +------------sibling------------+

        The problem I'd like to (or have to) address is that, if one of the proxies's direct connection is for some reason unavailable, the proxy must be able to be "functional" by querying its siblings. I guess in a longer term, is it possible for squid to be configured to dynamically adapt to its network connections?

============================
Gino LV. Ledesma
Campus Network Group
Ateneo de Manila University
http://cng.ateneo.net/
Received on Thu Sep 19 2002 - 01:34:58 MDT