On 2016-12-27 16:37, wei wrote:
The curl command is: curl -s -i http://showip.net [1] The configuration in squid.conf is: cache_peer 192.168.1.100 parent 3128 weight=5 no-query name=p1 round-robin cache_peer 192.168.1.101 parent 3128 weight=5 no-query name=p2 round-robin cache_peer 192.168.1.102 parent 3128 weight=5 no-query name=p3 round-robin
Since you have equal weightings on each peer they are irrelevant. Remove the weight=5 for better performance.
cache_peer_domain p1 .com cache_peer_domain p1 .net cache_peer_domain p2 .com cache_peer_domain p2 .net cache_peer_domain p3 .com cache_peer_domain p3 .net
Please use the cache_peer_access directive instead. cache_peer_domain is deprecated.
From: wei Hi Amos, The Squid version I tested is 3.4.14(installed by yum on Centos 6.8) and 3.4.8(installed by apt on Debian 7), both have the problem of use same parent proxy after about 55 seconds. I also test 3.5.23(compiled by default parameter) and found the time increased from 55 to 85 seconds or so. Thanks.
I have just checked the code. The round-robin counters are 'cleared' on a 5min cycle, on startup, on reconfigure, or when peers change from DEAD to ALIVE state "to prevent the revived CachePeer being flooded with requests which it has 'missed' during the down period."
So if your peers are being reset every minute, then either you are reconfiguring Squid that often, or they are constantly being detected as DEAD. Neither of which is good.
Amos _______________________________________________ squid-users mailing list squid-users@xxxxxxxxxxxxxxxxxxxxx http://lists.squid-cache.org/listinfo/squid-users
