On 18/11/2015 8:00 p.m., 徐永健 wrote: > Hi, Eliezer: > Thanks! > I just solved this problem.Things are as follows: > The hostname is S-LAB-53, and I saw WANINGS in cache.log. > ------------------------------------------------ > > 2015/11/18 11:16:52 kid1| WARNING: 'S-LAB-53' rDNS test failed: (2) No such file or directory > > 2015/11/18 11:16:52 kid1| WARNING: Could not determine this machines public hostname. Please configure one or set 'visible_hostname' > > -------------------------------------------------- > > So I changed '127.0.0.1 localhost' to '127.0.0.1 localhost S-LAB-53' in > '/etc/hosts' and restart squid. Then, the 2 seconds response time > problem gone. > > But the root cause is still unknown. :( Some requests just take time. In this case probably the time it takes for DNS to figure out that "S-LAB-53" is not a DNS registered Fully-Qualified Domain Name (FQDN). It is important that a proxy visible_hostname be configured with (or able to auto-detect) its public FQDN. Since that FQDN is used to construct URLs that get emitted to external clients in some messages. This happens in all proxy types, but has somewhat higher visibility in reverse-proxy / CDN. PS. I have been doing some updates in the client connection handling and managed yesterday to replicate the behaviour of seeing _ABORTED on every request. Do you happen to be patching your Squid in the client_side*.cc code files? if so the patch is probably broken/incorrect. If not, what is the output of "squid -v" from your proxy? Amos _______________________________________________ squid-users mailing list squid-users@xxxxxxxxxxxxxxxxxxxxx http://lists.squid-cache.org/listinfo/squid-users
