#!/bin/sh . ../.function DOC_ROOT=benchmark/docroot rm -f err/benchmarking.err \ benchmark/benchmarking.log* \ out/userver_*.out err/userver_*.err \ trace.*userver_*.[0-9]* object.*userver_*.[0-9]* stack.*userver_*.[0-9]* mempool.*userver_*.[0-9]* \ $DOC_ROOT/trace.*userver_*.[0-9]* $DOC_ROOT/object.*userver_*.[0-9]* $DOC_ROOT/stack.*userver_*.[0-9]* $DOC_ROOT/mempool.*userver_*.[0-9]* #UTRACE="0 50M 1" #UOBJDUMP="0 10M 100" #USIMERR="error.sim" #VALGRIND="valgrind -v --trace-children=yes" UMEMPOOL=135,0,53,134,1057,-24,-18,-21,43 export UTRACE UOBJDUMP USIMERR VALGRIND UMEMPOOL ORM_DRIVER="sqlite" ORM_OPTION="dbname=../db/hello_world" export ORM_DRIVER ORM_OPTION DIR_CMD="../../examples/userver" compile_usp ulimit -n 100000 echo 0 > /proc/sys/net/ipv4/tcp_syncookies # ----------------------------------------------------------------------------------------------------------------------------------------------------------- # cat /proc/sys/net/ipv4/tcp_mem # ------------------------------ # 47556 63410 95112 # ------------------------------------------- # low threshold => 47556 * 4096 = 185M # memory pressure => 63410 * 4096 = 247M # Out of socket memory => 95112 * 4096 = 371M # ------------------------------------------- # cat /proc/net/sockstat # cat /proc/sys/net/ipv4/tcp_max_orphans => 131072 # ----------------------------------------------------------------------------------------------------------------------------------------------------------- # So in this case we have 21564 orphans. That doesn't seem very close to 65536... Yet, if you look once more at the code above that prints the warning, # you'll see that there is this shift variable that has a value between 0 and 2, and that the check is testing (orphans << shift > sysctl_tcp_max_orphans). # What this means is that in certain cases, the kernel decides to penalize some sockets more, and it does so by multiplying the number of orphans by 2x # or 4x to artificially increase the "score" of the "bad socket" to penalize. The problem is that due to the way this is implemented, you can see a # worrisome "Out of socket memory" error when in fact you're still 4x below the limit and you just had a couple "bad sockets" (which happens frequently # when you have an Internet facing service). So unfortunately that means that you need to tune up the maximum number of orphan sockets even if you're # 2x or 4x away from the threshold. What value is reasonable for you depends on your situation at hand. Observe how the count of orphans in # /proc/net/sockstat is changing when your server is at peak traffic, multiply that value by 4, round it up a bit to have a nice value, and set it. # You can set it by doing a echo of the new value in /proc/sys/net/ipv4/tcp_max_orphans, and don't forget to update the value of net.ipv4.tcp_max_orphans # in /etc/sysctl.conf so that your change persists across reboots. # ----------------------------------------------------------------------------------------------------------------------------------------------------------- # A server that uses SYN cookies doesn't have to drop connections when its SYN queue fills up. # Instead it sends back a SYN+ACK, exactly as if the SYN queue had been larger. # (Exceptions: the server must reject TCP options such as large windows, and it must use one of the # eight MSS values that it can encode.) When the server receives an ACK, it checks that the secret # function works for a recent value of t, and then rebuilds the SYN queue entry from the encoded MSS. # ----------------------------------------------------------------------------------------------------------------------------------------------------------- #echo 1024 > /proc/sys/net/core/somaxconn #echo 1 > /proc/sys/net/ipv4/tcp_syncookies #echo 524288 > /proc/sys/net/ipv4/tcp_max_orphans #echo 2 > /proc/sys/net/ipv4/tcp_synack_retries # 5 -> 2 == 21 sec (Total time to keep half-open connections in the backlog queue) #echo 0 > /proc/sys/kernel/printk_ratelimit # 5 #echo 0 > /proc/sys/kernel/printk_ratelimit_burst # 10 # ----------------------------------------------------------------------------------------------------------------------------------------------------------- #STRACE=$TRUSS #VALGRIND="valgrind --tool=exp-dhat" #MUDFLAP_OPTIONS="-ignore-reads -backtrace=8" start_prg_background userver_tcp -c benchmark/benchmarking.cfg #start_prg_background userver_ssl -c benchmark/benchmarking_ssl.cfg #run command on another computer #ab -n 100000 -c10 http://stefano/servlet/benchmarking?name=stefano #ab -n 100000 -c10 https://stefano/servlet/benchmarking?name=stefano #$SLEEP #pkill userver_tcp userver_ssl mv err/userver_tcp.err err/benchmarking.err #mv err/userver_ssl.err err/benchmarking.err echo "PID = `cat benchmark/userver_tcp.pid`" # grep -v 'EAGAIN\|EPIPE\|ENOTCONN\|ECONNRESET' err/benchmarking.err # gprof -b ../../examples/userver/userver_tcp gmon.out >profile.out 2>/dev/null