On Dec 11, 2007, at 2:02 PM, Ricardo Newbery wrote:
From my reading and testing, it appears that when ICP is enabled,
the peer is always selected based on the round trip time. There
doesn't seem to be a way to use something like round-robin load
balancing with ICP to drop problem peers. Is this true?
If so, then is there any other way to drop problem peers from the
round-robin pool?
Note, this is a reverse-proxy configuration.
Ric
Okay, found the answer on irc, I think.
Use the monitorurl or connect-timout option in cache_peer to drop bad
peers. Can anyone confirm?
Thanks,
Ric
