Cacher is the code snippet organizer for pro developers

命令

perf record --call-graph dwarf -e cpu-clock -F 99 -a -g -p $(ps aux | grep DataService | grep -v grep | awk '{print $2}') -- sleep 120
perf script -i perf.data | FlameGraph/stackcollapse-perf.pl | FlameGraph/flamegraph.pl > process.svg

diff flame

perf record -F 99 -a -g -- sleep 30
perf script > out.stacks1

perf record -F 99 -a -g -- sleep 30
perf script > out.stacks2

cd FlameGraph
./stackcollapse-perf.pl ../out.stacks1 > out.folded1
./stackcollapse-perf.pl ../out.stacks2 > out.folded2
./difffolded.pl out.folded1 out.folded2 | ./flamegraph.pl > diff2.svg

准备

perf-tools: https://github.com/brendangregg/perf-tools

FlameGraph: https://github.com/brendangregg/FlameGraph

perf example: http://www.brendangregg.com/perf.html#Symbols

• 用 -fno-omit-frame-pointer 重新编译
• --call-graph dwarf 视情况而定，使用错误可能会 No stack

-fno-omit-frame-pointer 避免优化帧指针，能够更好的还原调用堆栈。-O2 级别以下的优化选项默认带

Q&A

火焰图采样少

• 频率低，增加 -F 参数大小
• 时间短，增加记录时长
• 线程名类似 #11