User manual

Chapter 3

software manages its own disk storage by removing old,

infrequently-requested documents when the available disk space

is low.

Remove Cache Statistics Now. This option removes all log ﬁles

and statistics ﬁles stored on the CacheRaQ.

Cache Cluster Settings. This option lets you conﬁgure the

CacheRaQ as a Cluster Master. To do this, you simply enter the

host names (or IP addresses) of Cluster Slaves in the ﬁelds

marked “Cluster Slave 1,” “Cluster Slave 2,” etc. Make sure that

the speciﬁed Cluster Slaves are on the network and are up and

running. Also make sure that the admin password on the Cluster

Slaves is the same as the admin password on the Cluster Master.

This is necessary for the Cluster Master to retrieve statistics data

from the Cluster Slaves. For optimal performance, the Cluster

Slaves should be conﬁgured for Traditional Proxy Caching only

(that is, do not enable Transparent Caching on the Cluster

Slaves).

The Cluster Master can operate in any caching mode —

Traditional Proxy Caching mode or Transparent mode with 1 or 2

network interfaces. In Transparent mode, all client trafﬁc should

be routed through the Cluster Master, just as if the Cluster Master

were a standalone CacheRaQ operating in Transparent mode.

Inter-Cache Protocol (ICP) Settings. The Inter-Cache Protocol

(ICP) is an Internet protocol used to create cache hierarchies. The

ICP Conﬁguration page allows you to conﬁgure ICP peer hosts

for the CacheRaQ. You may wish to conﬁgure ICP if, for example,

your Internet Service Provider (ISP) operates an ICP-compliant

cache with which you want to peer. On the IPC Conﬁguration

page, each peer requires you to specify the following

information: the peer host name, an indication of whether the

peer is a “parent” or a “sibling” (in ICP terminology), and HTTP

and ICP port numbers to use for the speciﬁed peer. Optionally,

you can conﬁgure the peer to be “Proxy-only,” in which case

documents retrieved from this peer will be returned to clients but

will not be stored locally by the Cluster Master. Additionally, you

can conﬁgure the peer as “No-query”, in which case cache misses

are automatically forwarded to the conﬁgured peer.

Cache Clustering can be used in conjunction with ICP in the

following way: Conﬁgure the Cluster Master as described in

“Cache Cluster Settings,” above, but do not conﬁgure any ICP

peers for it. On the Cluster Slaves, conﬁgure the desired ICP