分析工具集合
strace
netstat
perf
dtrace
stap
iotop
blktrace
top
pidstat
mpstat
dstat
vmstat
ping
tcpdump
ip
nicstat
slabtop
free
sar
/proc
基本工具
-
uptime: show load averages(1min, 5mins, 15mins), no more than cpu count
-
top: system-wide and per-process summaries
-
mpstat: check for hot threads, unbalanced workoads
mpstat -P ALL 1
-
iostat: 磁盘IO统计
iostat -xkdz 1
-
vmstat: 虚拟内存使用统计
vmstat 1
-
free: 内存使用情况统计(KB)
-
ping: 测网络时延
-
dstat: 比vmstat更友好，彩色
中级工具
-
sar: system activity reporter
pageing -B; block device -d; run queue -q
-
netstat: various network protocol statistics using -s
-
pidstat: very useful process breakdowns
pidstat -d l disk io
-
strace: system call tracer
strace -tttT -p pid
“high application latency often caused by resource io, and most resource io is performed by syscalls”
-
tcpdump: 嗅探网络包，dump到文件
tcpdump -i eth4 -w /tmp/out.tcpdump tcpdump -nr /tmp/out.tcpdump -
blktrace: tracing块设备IO事件
-
iotop: 进程的磁盘IO
iotop -bod5
-
slabtop: kernel slab allocator usage top
slabtop -sc # shows where kerrnel momory is consumed
-
sysctl: 系统设置
-
/proc: read statistic sources directly
高级工具
-
perf: performance counters
perf stat gzip file1
-
dtrace: programmable, real-time, dynamic and static tracing
dtrace -n 'provider:module:function:name /predicate/ {action}' probe description filter -
systemtap: static and dynamic tracing, probes, tapsets, scripts,…