Hi, folks, My squid box has terminated service twice for these two days. Before this, it was running pretty well for about two weeks without any config change. It serves that load about 200 req/s (peak), and 130 req/s (mean).
--------- System informatio ------------ - AMD Opteron 248 * 2 - S2882 Thunder K8s Pro - RAM 1GB - 2 SCSI Seagate 10k for RAID-0 # uname -a Linux NGate 2.4.21-20.EL.NGate #2 SMP Mon Nov 8 13:26:37 CST 2004 i686 athlon i386 GNU/Linux # /usr/local/squid/sbin/squid -v Squid Cache: Version 2.5.STABLE7 configure options: --prefix=/usr/local/squid --with-aufs-threads=32 --with-pthreads --with-aio --with-dl --enable-storeio=ufs,aufs,diskd --enable-removal-policies=lru,heap --enable-kill-parent-hack --enable-snmp --enable-poll --disable-ident-lookups --disable-hostname-checks --enable-underscores --enable-stacktraces --enable-dl-malloc --enable-wccpv2 # cat /usr/local/squid/etc/squid.conf visible_hostname NGate.com hierarchy_stoplist cgi-bin ? acl QUERY urlpath_regex cgi-bin \? .cgi .pl .php .asp .cfm no_cache deny QUERY refresh_pattern ^ftp: 1440 20% 10080 refresh_pattern ^gopher: 1440 0% 1440 refresh_pattern -i .gif 1440 90% 129600 reload-into-ims refresh_pattern -i .swf 1440 90% 129600 reload-into-ims refresh_pattern -i .jpg 1440 90% 129600 reload-into-ims refresh_pattern -i .bmp 1440 90% 129600 reload-into-ims refresh_pattern -i .pdf 0 90% 129600 reload-into-ims refresh_pattern -i .zip 0 90% 129600 reload-into-ims refresh_pattern -i .rar 0 90% 129600 reload-into-ims refresh_pattern -i .exe 0 90% 129600 reload-into-ims refresh_pattern . 1 20% 4320 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 SSL_ports port 443 acl Safe_ports port 80 acl CONNECT method CONNECT http_access allow manager localhost http_access allow CONNECT SSL_ports http_access deny !Safe_ports http_access deny to_localhost http_access allow all acl snmppublic snmp_community public snmp_access allow snmppublic all snmp_incoming_address 0.0.0.0 snmp_outgoing_address 0.0.0.0 dns_nameservers 127.0.0.1 half_closed_clients off logfile_rotate 0 maximum_object_size 51200 KB minimum_object_size 0 KB maximum_object_size_in_memory 100 KB store_avg_object_size 13 KB store_objects_per_bucket 128 cache_dir aufs /cache 30720 128 256 cache_mem 128 MB cache_replacement_policy heap LFUDA memory_replacement_policy heap LRU coredump_dir /cache cache_access_log /var/log/cache/access.log cache_log /var/log/cache/cache.log cache_store_log none forwarded_for off httpd_accel_host virtual httpd_accel_port 80 httpd_accel_with_proxy on httpd_accel_uses_host_header on httpd_accel_single_host off http_port 3333 icp_port 0 snmp_port 3401 wccp_router xxx.xxx.xxx.xxx And I got something in /var/log/messages Nov 23 03:20:02 NGate kernel: Mem-info: Nov 23 03:20:02 NGate kernel: Zone:DMA freepages: 1055 min: 1056 low: 1088 high: 1120 Nov 23 03:20:02 NGate kernel: Zone:Normal freepages: 702 min: 703 low: 3968 high: 5728 Nov 23 03:20:02 NGate kernel: Zone:HighMem freepages: 0 min: 0 low: 0 high: 0 Nov 23 03:20:02 NGate kernel: Free pages: 1757 ( 0 HighMem) Nov 23 03:20:02 NGate kernel: ( Active: 216824/176, inactive_laundry: 465, inactive_clean: 0, free: 1757 ) Nov 23 03:20:02 NGate kernel: aa:1666 ac:14 id:3 il:0 ic:0 fr:1055 Nov 23 03:20:02 NGate kernel: aa:213121 ac:2113 id:547 il:0 ic:0 fr:702 Nov 23 03:20:02 NGate kernel: aa:0 ac:0 id:0 il:0 ic:0 fr:0 Nov 23 03:20:02 NGate kernel: 1*4kB 1*8kB 1*16kB 1*32kB 1*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 1*4096kB = 4220kB) Nov 23 03:20:02 NGate kernel: 2*4kB 2*8kB 0*16kB 9*32kB 1*64kB 1*128kB 1*256kB 0*512kB 0*1024kB 1*2048kB 0*4096kB = 2808kB) Nov 23 03:20:02 NGate kernel: Swap cache: add 8508114, delete 8508114, find 4226233/6345266, race 0+25 Nov 23 03:20:02 NGate kernel: 2992 pages of slabcache Nov 23 03:20:03 NGate kernel: 256 pages of kernel stacks Nov 23 03:20:03 NGate kernel: 677 lowmem pagetables, 0 highmem pagetables Nov 23 03:20:03 NGate kernel: Free swap: 0kB Nov 23 03:20:03 NGate kernel: 229376 pages of RAM Nov 23 03:20:03 NGate kernel: 0 pages of HIGHMEM Nov 23 03:20:03 NGate kernel: 4702 reserved pages Nov 23 03:20:03 NGate kernel: 2671 pages shared Nov 23 03:20:03 NGate kernel: 0 pages swap cached Nov 23 03:20:03 NGate kernel: Out of Memory: Killed process 16482 (squid). Nov 23 03:20:03 NGate kernel: Out of Memory: Killed process 16484 (squid). Nov 23 03:20:03 NGate kernel: Out of Memory: Killed process 16485 (squid). Nov 23 03:20:03 NGate kernel: Out of Memory: Killed process 16486 (squid). I've read some releative sections in squid FAQ, it told me that "as the software has matured, we believe almost all of Squid's memory leaks have been eliminated, and new ones are least easy to identify." So, this can't be memory leak? Maybe something wrong in my squid.conf? Is maximum_object_size and cache_mem a bit too high?
