Napp-IT Benchmark

Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.

Rand__

Well-Known Member
Mar 6, 2014
6,626
1,767
113
So trying to run some benches comparing to the benchmark pdf.

Differences to the documentation I've found/ I cannot recreate identically (newest release of napp-it/pdf)

1. bench1.sh
I assume that needs to be prepared by creating a bench1.sh containing the command provided (which has a typo in the location) ("/var(web-gui")

2. There is no filebench + read + write option?
But I see that read and write is now separate so I assume this is the equivalent?

Manual:
upload_2018-2-25_20-9-53.png

Current Gui:
upload_2018-2-25_21-52-23.png

Whats the expected duration for this?


begin test 9 singlestreamread.f .. is running for 90 minutes quite a long time now...

root@napp-it-san024-1801:/var/web-gui/_log/tunings# ps -ef |grep -s 24982
root 15853 16609 0 20:22:17 pts/1 0:00 grep -s 24982
root 24982 1 0 18:33:20 ? 14:03 seqread -a seqread -i 1 -s f6a00000 -m /tmp/filebench-shm-K1aGTW

root@napp-it-san024-1801:/var/web-gui/_log/tunings# ps -ef |grep -s 24982
root 18351 16609 0 21:46:36 pts/1 0:00 grep -s 24982
root 24982 1 6 18:33:20 ? 24:59 seqread -a seqread -i 1 -s f6a00000 -m /tmp/filebench-shm-K1aGTW

Rebooted to get a clean state after 3hrs, webgui had timed out a while ago...



Realized that was the wrong test;) Test 4 & 9 need to be changed from the default value
 
Last edited:

gea

Well-Known Member
Dec 31, 2010
3,141
1,182
113
DE
A shellscript like bench1.sh is only an option if you want to modify systemparameter prior a benchmark ex for a tuning series with different system settings.

Software is ongoing development where I change defaults to change the benchmarks for a better understanding of expected storage behaviour. A test sequence like the pdf is a snapshot where I care of same hardware and workloads. Another hardware or workload series will give different results.

The defaults are for a fast storage system. On a slower system you time out on tests. In this case select either a faster dd for read or write or select a partly filemicro workload that runs very short (often tests RAM performance only) to get a wanted read or write workload finished.
 

Rand__

Well-Known Member
Mar 6, 2014
6,626
1,767
113
Well I wanted to get comparable values so thats why I tried to mimic your example.
 

gea

Well-Known Member
Dec 31, 2010
3,141
1,182
113
DE
Years ago my prof told me "Wer misst misst Mist" (who measures measures crap) so benchmarks are only exact within the test environment.

In my pdf I compared different disk types with different slogs or vdevs and different RAM sizes on a given hardware. The same series on a different hardware will give other results where I only expect a similar conclusion from the benchmarks.

If you use the same benchmarks/ filebench workloads on a similar hardware, you should get similar values, otherwise you can get the result "within expected or outside expected" values.