Date: Mon, 27 Nov 2006 10:44:58 +0300 From: =?koi8-r?Q?=F2=C9=C8=C1=C4_=E7=C1=C4=D6=C9=C5=D7?= <rihad@mail.ru> To: freebsd-questions@freebsd.org Subject: zoneli Message-ID: <E1GobAg-000M1e-00.rihad-mail-ru@f44.mail.ru>
next in thread | raw e-mail | index | archive | help
Hi! From time to time our moderately loaded Squid 2.6.3 (~500 simultaneous clients, ~5000 req/min) stops responding all of a sudden, its status being 'zoneli' according to top(1), and nothing short of a machine reboot can bring it back to life. I think it's the growing load on the box that causes this, because the problem never showed up when the load was lower. I upped certain kernel variables through /boot/loader.conf, it seemed to help in the sense that zoneli now occurs much less frequently (like once in a month). Can you please give any suggestions on how to fix the problem? If you should need any further info, please let me know. $ uname -r 6.0-RELEASE-p15 $ grep '^CPU:' /var/run/dmesg.boot CPU: AMD Athlon(tm) 64 Processor 3000+ (1808.81-MHz 686-class CPU) $ grep -E '^(real|avail) memory' /var/run/dmesg.boot real memory = 2147418112 (2047 MB) avail memory = 2097672192 (2000 MB) $ grep -v -e ^# -e '^$' /boot/loader.conf kern.maxdsiz="1153433600" kern.maxssiz="201326592" kern.ipc.nmbclusters="65536" kern.ipc.msgmnb="8192" kern.ipc.msgssz="64" kern.ipc.msgtql="2048" $ grep -v -e ^# -e '^$' /etc/sysctl.conf net.inet.tcp.sendspace=131072 net.inet.tcp.recvspace=131072 kern.ipc.somaxconn=1024 kern.ipc.maxsockbuf=524288
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?E1GobAg-000M1e-00.rihad-mail-ru>