Extract thoughput value from JMeter log file

Question

I use JMeter for load testing and want to report error on CI pipeline when application though put is less than 1/sec. JMeter logs summary of results as -

2017/01/17 01:57:00 INFO  - jmeter.reporters.Summariser: summary +    103 in 00:00:30 =    3.4/s Avg:  2937 Min:    48 Max:  7962 Err:     0 (0.00%) Active: 2 Started: 10 Finished: 8 
2017/01/17 01:57:00 INFO  - jmeter.reporters.Summariser: summary =  12218 in 01:00:03 =    3.4/s Avg:  2942 Min:    20 Max: 27317 Err:     1 (0.01%)
2017/01/17 01:57:00 INFO  - jmeter.reporters.Summariser: summary +      1 in 00:00:00 =   66.7/s Avg:  3227 Min:  3227 Max:  3227 Err:     0 (0.00%) Active: 0 Started: 10 Finished: 10 
2017/01/17 01:57:00 INFO  - jmeter.reporters.Summariser: summary =  12219 in 01:00:03 =    3.4/s Avg:  2942 Min:    20 Max: 27317 Err:     1 (0.01%)  

Herein I would capture the last line and if value of throughput is less than 1 (which is 3.4/s in the current example) then it would be reported as error. Here is the script I wrote to capture value of throughput -

grep "meter.reporters.Summariser: summary" test.log | tail -1 | awk '{print $12}' | cut -d "." -f1

and it gets me 3 which is ok. How can I improve this script?

Answer 1

You are using here four different tools: grep, tail, awk and cut to do the following:

• select lines
• get the last one
• print a given field of that one
• extract the integer part of that field

This is clear and it works. However, using four processes is a bit of overkill, since all of these can be performed one-handedly via awk:

awk '/meter.reporters.Summariser: summary/ {rec=$12} END{printf "%d\n", rec}' file

• /meter.reporters.Summariser: summary/ {rec=$12}
  When reading from the file, /pattern/ will match those lines containing the "pattern". If this happens, the action within {} is performed. In this case, it consists in storing the 12^th field.
  But wait, weren't we looking for the last line in which this occurs? Yes, and this is exactly what we are doing by overwriting rec every time the pattern happens: last time it is written, it will store the definitive result.

• END{printf "%d\n", rec}
  Once all lines have been processed, red will have the desired value on the form 3.4/s. We just want the 3 to appear, so we could do some string triming similar to the one you did with cut. However, we can also use awk's printf() to use Format-Control Letters to just print the integer part of it. This will also print 0 if rec is empty:

  $ awk 'BEGIN {printf "%s\n", "3.4/s"}'
  3.4/s
  $ awk 'BEGIN {printf "%d\n", "3.4/s"}'
  3
  $ awk 'BEGIN {printf "%d\n", ""}'
  0
  

Of course, you could also add the threshold logic in the script.