Hi all -
I've been playing around with Squid 3.0.PRE7 a bit and have a small
question on using (weighted) round-robin.
This is in a reverse proxy setting, naturally.
I have a cluster of back-end machines set up using cache_peer and
weighted round robin:
cache_peer [IP] parent 8080 0 no-query originserver weighted-round-
robin name=server1
cache_peer [IP] parent 8080 0 no-query originserver weighted-round-
robin name=server2
I have some other machines that handle computationally intensive
tasks. I want to send specific requests to these boxes based on url:
cache_peer [IP] parent 8080 0 no-query originserver name=box1
I have (I think) appropriate acls to access these back end machines
and am using cache_peer_access to drive requests to specific boxes:
acl server1 dstdomain server1
acl server2 dstdomain server2
acl ComputationalProcess urlregex_path -i bigjob
cache_peer_access ComputationalProcess allow box1
cache_peer_access deny server1 server2
Two questions:
1. Squid doesn't seem to be able to choose the box1 originserver
unless it is also included in the round-robin pool. Is this true /
the correct way to do this?
2. For each job I have, I have to specifically set lets of allow/deny
cache_peer_access lines. Is this perhaps best handled as a
url_rewrite_program instead?
Thanks!
Todd