G'day,
I've been tweaking my squids today and morning the loss of
single_parent_bypass in 2.x squids, when I thought I'd try it anyway. It
seems that squid does not complain if you put one in, but it just ignores
it. Is this the case?
I would really like some sort of return to single_parent_bypass behaviour.
I have a complex series of upstream proxies where combinations of
cache_peer_access acl's mean that only a single parent is available for a
about 50% of requests, but the other 50% of requests use ICP to pick the
best parent. Something like single_parent_bypass would save me significant
latency and ICP traffic. AFAIKT, there is no way to configure 2.x squids to
do this.
I also recently had a need to have a peer behave like a parent at certain
times, and a sibling at others. Are there any plans to extend
neighbor_type_domain using acl's? IMHO, it would be good if all the options
like this were changed to use ACL's. In the mean time, any suggestions as to
the best way to do this?
-- ---------------------------------------------------------------------- ABO: finger abo@minkirri.apana.org.au for more info, including pgp key ----------------------------------------------------------------------Received on Wed Jul 31 2002 - 02:23:11 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 17:09:24 MST