On Wed, 7 Dec 2005, Mark Elsen wrote:
For various reasons I was given a Sun v20z with Solaris 10 x86 to run
Squid on. But when testing (with small and heavy load) I get the following
in cache.log:
2005/12/07 11:15:05| clientReadRequest: FD 15: no data to process ((11)
Resource temporarily unavailable)
2005/12/07 11:15:05| clientReadRequest: FD 20: no data to process ((11)
Resource temporarily unavailable)
and one cpu spins at 100% load and the proxy grinds to a halt.
...
This indicates a kernel resource problem.
Some network buffer may need increasing (e.g.)
Checkout the system (error) logs too; you may see
additional informational errors concerning this issue.
I looked through dmesg and /var/adm/messages but found no hints.
I tried boosting tcp parameters too but no effect.
ndd -set /dev/tcp tcp_conn_req_max_q 512
ndd -set /dev/tcp tcp_conn_req_max_q0 10240
ndd -set /dev/tcp tcp_rexmit_interval_initial 3000
ndd -set /dev/tcp tcp_rexmit_interval_min 2000
ndd -set /dev/tcp tcp_ip_abort_interval 600000
ndd -set /dev/tcp tcp_ip_abort_cinterval 60000
ndd -set /dev/tcp tcp_rexmit_interval_max 60000
ndd -set /dev/tcp tcp_keepalive_interval 3600000
ndd -set /dev/tcp tcp_fin_wait_2_flush_interval 67500
ndd -set /dev/tcp tcp_time_wait_interval 60000
ndd -set /dev/ip ip_icmp_err_interval 0
ndd -set /dev/ip ip_forward_src_routed 0
ndd -set /dev/tcp tcp_smallest_anon_port 8192
ndd -set /dev/udp udp_smallest_anon_port 8192
ndd -set /dev/tcp tcp_xmit_hiwat 32768
Machine have 2GB RAM and 2 cpu so is clearly enough to run squid.
"ulimit -Hn" reports 65536, so there is plenty of file descriptors.
Stumped :/
