Ouput Explanation of lsperfm - logstash performance testing


#1

Hi,

I am trying to use lsperfm, a logstash peformance testing tool. I have followed their installation instruction as mentioned in the README.md in their github wiki. I have logstash 1.5.2 installed in logstash 1.5.2/.

I faced a problem in the bootstrap step. After executing rake bootstrap and lsperfm-deps when I tried to execute bin/logstash deps it threw me an error No such command "deps". Other than this the installation went fine.

On executing lsperm from my logstash root directory. I got the below output:-

.................................................................................................................................................................................................................................................................................................................................simple line in/out,7.94,120.33,2688000,22339,28700.00,27043
simple line in/json out,13.03,120.32,2682000,22291,29600.00,27300
json codec in/out,8.65,120.04,1896000,15795,19500.00,18652
line in/json filter/json out,7.96,120.18,1398000,11633,17200.00,14226
apache in/json out,8.37,120.08,2372040,19755,28000.00,26260
apache in/grok codec/json out,7.82,120.11,1051380,8753,11000.00,10365
syslog in/json out,9.14,120.81,294000,2434,3000.00,2882

Can some please explain me this output? What does each of this comma-seperated values signify?


(Pere Urbon-Bayes) #2

Thanks for testing this gem, however I must tell you nowadays this is still experimental, so things can change a lot till version 1.0 :smile:.

Regarding the installation procedures, this part is completely misleading, thanks for noticing the bug. bin/logstash deps is only for versions before 1.5. This is why you are able to run the test without problems.

Regarding the output, this is a bug, you should see explanatory headers, but the output is missing them. I will explain, briefly what they are here in order as they appear:

  • The name of the test.
  • The time necessary to bootstrap/start logstash for this test.
  • How long the test was running.
  • Number of events processed.
  • Average events per second during the test.
  • Best events per second during the test.
  • Top 20% average events per second during the test.

As you could imagine this is basically feeding different data and different configurations to Logstash to then get some numbers out. You can see the configurations used inside the defaults directory.

/cheers

  • purbon

#3

Hi Purbon,

I have a doubt regarding the "The time necessary to bootstrap/start logstash for this test." (bTime), "How long the test was running" (exeTime) and the "test duration" in the suites(dur).

  1. Does dur include bTime as well or the test runs for dur after doing the bootstrapping for bTIme?
  2. What does exeTime signify? I have seen dur is not equal to exeTime some times.

(Vachan D A) #4

How to use this tool with custom logstash configuration, which reads from file for input?


#5

Mention the input file in your suites. Please refer a sample suite here which reads from a input file named catalog.txt placed under input folder.


(system) #6