Hi I'm using Squid in accelerator mode for a fairly busy Web site (ca. 200-250 req/s in peak). Everything is currently hosted on the same server - Squid, Apache and MySQL, but the machine is powerful enough (2xXeon 3,2GHz, 2GB RAM) to handle the load (50% idle, no swap used, 1GB cached, 300MB for buffers). I have around 60% hit rate for requests - all static content is cached in memory, and dynamic pages are forwarded to Apache. I have a problem with Squid not accepting connecting in a timely manner. Here is a tcpdump snippet: 16:38:37.240306 IP client.42054 > squid.80: S 487460662:487460662(0) win 5840 <mss 1460,nop,nop,timestamp 2904533217 0,nop,wscale 0> 16:38:40.239044 IP client.42054 > squid.80: S 487460662:487460662(0) win 5840 <mss 1460,nop,nop,timestamp 2904536217 0,nop,wscale 0> 16:38:46.237981 IP client.42054 > squid.80: S 487460662:487460662(0) win 5840 <mss 1460,nop,nop,timestamp 2904542217 0,nop,wscale 0> 16:38:58.233859 IP client.42054 > squid.80: S 487460662:487460662(0) win 5840 <mss 1460,nop,nop,timestamp 2904554217 0,nop,wscale 0> 16:38:58.233872 IP squid.80 > client.42054: S 83236182:83236182(0) ack 487460663 win 5792 <mss 1460,nop,nop,timestamp 402554534 2904554217,nop,wscale 0> 16:38:58.233973 IP client.42054 > squid.80: . ack 1 win 5840 <nop,nop,timestamp 2904554217 402554534> client and squid are connected to the same Gigabit Ethernet switch, so there is no problem with connectivity between them. What can cause such behavior? Is it a problem with Squid itself, Squid configuration or maybe kernel configuration? I'm running Debian Linux sarge, kernel 2.6.8-2-686-smp with /proc/sys/net/ipv4/tcp_max_syn_backlog=4096. My squid.conf: http_port ip1:80 http_port ip2:80 icp_port 0 hierarchy_stoplist cgi-bin ? acl QUERY urlpath_regex cgi-bin no_cache deny QUERY cache_mem 256 MB cache_dir ufs /var/spool/squid 512 16 256 cache_store_log none debug_options ALL,1 hosts_file /etc/hosts redirect_rewrites_host_header off refresh_pattern . 0 20% 4320 read_timeout 30 seconds request_timeout 30 seconds persistent_request_timeout 10 seconds half_closed_clients off pconn_timeout 120 seconds acl all src 0.0.0.0/0.0.0.0 acl manager proto cache_object acl localhost src 127.0.0.1/255.255.255.255 acl to_localhost dst 127.0.0.0/8 acl purge method PURGE acl CONNECT method CONNECT http_access deny blokuj http_access allow manager localhost http_access deny manager http_access allow purge localhost http_access deny purge http_access deny !Safe_ports http_access deny CONNECT !SSL_ports http_access allow all http_reply_access allow all icp_access deny all httpd_accel_host 127.0.0.1 httpd_accel_port 80 httpd_accel_single_host on httpd_accel_with_proxy off httpd_accel_uses_host_header on httpd_accel_no_pmtu_disc on buffered_logs on coredump_dir /var/spool/squid client_persistent_connections off server_persistent_connections on pipeline_prefetch on Any help would be greatly appreciated. Best regards, Michal Gruchala
