performance / tuning tips. to the point.
|
Analyzing Network Performance
Linux Windows AIX Solaris
Linux
Use 'netstat' with -p option to display PID/Program name for sockets:
# netstat -p 2
Active Internet connections (w/o servers)
Proto Recv-Q Send-Q Local Address Foreign Address State PID/Program name
tcp 1 0 pw101.perfwiki.com:56451 pw101.perfwiki.com:9081 CLOSE_WAIT 31901/httpd
tcp 1 0 pw101.perfwiki.com:56460 pw101.perfwiki.com:9081 CLOSE_WAIT 32008/httpd
tcp 1 0 pw101.perfwiki.com:56456 pw101.perfwiki.com:9081 CLOSE_WAIT 32006/httpd
tcp 1 0 pw101.perfwiki.com:56446 pw101.perfwiki.com:9081 CLOSE_WAIT 31898/httpd
tcp 1 0 pw101.perfwiki.com:56447 pw101.perfwiki.com:9081 CLOSE_WAIT 31899/httpd
tcp 1 0 pw101.perfwiki.com:56445 pw101.perfwiki.com:9081 CLOSE_WAIT 31897/httpd
tcp 1 0 pw101.perfwiki.com:35518 pw101.perfwiki.com:9081 CLOSE_WAIT 31900/httpd
tcp 0 196 pw101.perfwiki.com:ssh 9.65.243.110:2708 ESTABLISHED 28120/0
tcp 0 0 pw101.perfwiki.com:56437 legolas.stl.ib:DB2_db2inst1 ESTABLISHED 31903/java
Active UNIX domain sockets (w/o servers)
Proto RefCnt Flags Type State I-Node PID/Program name Path
unix 6 [ ] DGRAM 1879 1766/syslogd /dev/log
unix 3 [ ] STREAM CONNECTED 575699094 31983/db2wdog
unix 3 [ ] STREAM CONNECTED 575699093 31983/db2wdog
unix 3 [ ] STREAM CONNECTED 575699092 31983/db2wdog
unix 3 [ ] STREAM CONNECTED 575699091 31983/db2wdog
unix 3 [ ] STREAM CONNECTED 575699090 31983/db2wdog
unix 3 [ ] STREAM CONNECTED 575699089 31983/db2wdog
unix 3 [ ] STREAM CONNECTED 575699088 31983/db2wdog
unix 3 [ ] STREAM CONNECTED 575699087 31983/db2wdog
unix 3 [ ] STREAM CONNECTED 575699086 31983/db2wdog
unix 3 [ ] STREAM CONNECTED 575699085 31983/db2wdog
...
Looks like db2wdog is a major program using the network (it has active sockets).
|
Windows
'Perfmon' is a Windows tool that can be configured to monitor (and log to
file) system resources. Physical disk counters are one area that can be
added to the overall resources to be monitored. You can start 'perfmon'
simply by typing 'perfmon' in the "Start->Run..." window:
AIX
A very useful command called 'netpmon' is available on AIX. It uses
'trace' to monitor activity and report statistics on network I/O usage and
network-related CPU usage. One can easily determine that if a process is
causing high network-related CPU usage, chances are it is causing a lot of
network traffice.
# netpmon -o netmon.out; sleep 10; trcstop
# more netmon.out
Sun Dec 4 19:55:40 2005
System: AIX pw101 Node: 5 Machine: 002FBF7D4C00
========================================================================
Process CPU Usage Statistics:
-----------------------------
Network
Process (top 20) PID CPU Time CPU % CPU %
----------------------------------------------------------
rcp 349885 0.7986 0.660 0.522
db2fm 512122 0.0375 0.073 0.000
rcp 349882 0.2821 0.234 0.171
netpmon 249960 0.0225 0.044 0.000
db2fm 1962022 0.0136 0.026 0.000
db2set 516208 0.0096 0.019 0.000
java 1716352 0.0070 0.014 0.000
db2disp 1143036 0.0052 0.010 0.000
swapper 0 0.0040 0.008 0.000
trcstop 512124 0.0036 0.007 0.000
UNKNOWN 647338 0.0025 0.005 0.000
db2fmcd 311516 0.0022 0.004 0.000
java 536650 0.0019 0.004 0.000
getty 290960 0.0019 0.004 0.000
db2hmon 1458184 0.0014 0.003 0.000
gil 77862 0.0014 0.003 0.003
sshd: 737282 0.0014 0.003 0.000
ksh 2027638 0.0012 0.002 0.000
java 331974 0.0008 0.001 0.000
netpmon 1470576 0.0007 0.001 0.000
sddsrv 266388 0.0006 0.001 0.000
sched 12294 0.0004 0.001 0.000
----------------------------------------------------------
Total (all processes) 0.1216 0.236 0.696
The "Network CPU%" column is the percentage of total time that this process spent executing
network-related code.
|
Solaris
Use 'netstat' and the following script to determine the PID that is
using port(s) which causing the highest network throughput:
#!/bin/ksh
#
# Orf Gelbrich
# 7-30-2003
# find from a port the pid that started the port
#
# Need to be root to check other folk's PID's out
#
line='-------------------------------------------------------------------------'
pids=$(/usr/bin/ps -ef | sed 1d | awk '{print $2}')
# Prompt users or use 1st cmdline argument
if [ $# -eq 0 ]; then
read ans?"Enter port you like to know pid for: "
else
ans=$1
fi
# Check all pids for this port, then list that process
for f in $pids
do
/usr/proc/bin/pfiles $f 2>/dev/null | /usr/xpg4/bin/grep -q "port: $ans"
if [ $? -eq 0 ] ; then
echo $line
echo "Port: $ans is being used by PID:\c"
/usr/bin/ps -ef -o pid -o args | egrep -v "grep|pfiles" | grep $f
fi
done
exit 0
** from http://sysunconfig.net/unixtips/solaris.html
|
|
|
|
