|
| 1 | +acrntrace |
| 2 | +######### |
| 3 | + |
| 4 | +DESCRIPTION |
| 5 | +########### |
| 6 | +acrntrace: is a tool running on SOS, to capture trace data. |
| 7 | +scripts directory includes scripts to analyze the trace data. |
| 8 | + |
| 9 | +USAGE |
| 10 | +##### |
| 11 | +Capture trace data on SOS |
| 12 | + |
| 13 | + 1) Launch acrntrace |
| 14 | + Capture buffered trace data: |
| 15 | + # acrntrace |
| 16 | + or clear buffer before tracing start: |
| 17 | + # acrntrace -c |
| 18 | + Trace files are created under /tmp/acrntrace/, directory name with time |
| 19 | + string eg: 20171115-101605 |
| 20 | + 2) To stop acrntrace |
| 21 | + # q <enter> |
| 22 | + 3) Copy the trace data to linux pc |
| 23 | + # scp -r /tmp/acrntrace/20171115-101605/ xxx@10.239.142.239:/home/xxxx/t |
| 24 | + race_data |
| 25 | + |
| 26 | +Analyze the trace data on Linux PC |
| 27 | + |
| 28 | + 1) Run the python script to analyze the vm_exits: |
| 29 | + # acrnalyze.py -i /home/xxxx/trace_data/20171115-101605/0 -o /home/xxxx/trac |
| 30 | + e_data/20171115-101605/cpu0 --vm_exit |
| 31 | + - "--vm_exit" specify the analysis to do, currently, only vm_exit analysis |
| 32 | + is supported. |
| 33 | + - A preprocess would be taken out to make the trace data start and end with |
| 34 | + an VM_ENTER, and a copy of original data file is saved with suffix ".orig"; |
| 35 | + - Analysis report would be given on the std output and in a csv file with |
| 36 | + name specified via "-o outpu_file"; |
| 37 | + Script usage: |
| 38 | + [Usage] acrnalyze.py [options] [value] ... |
| 39 | + [options] |
| 40 | + -h: print this message |
| 41 | + -i, --ifile=[string]: input file |
| 42 | + -o, --ofile=[string]: output file |
| 43 | + --vm_exit: to generate vm_exit report |
| 44 | + |
| 45 | + The scripts require bash and python2. |
| 46 | + |
| 47 | +BUILD |
| 48 | +##### |
| 49 | +# make |
0 commit comments