Replace dd with fio? about yet-another-bench-script HOT 5 CLOSED

I agree that dd (and ioping for that matter) are both inadequate measures of disk performance. I did try to get fio working with this, however it'll either require elevated privileges (to install fio) or the user to install prerequisites (building fio from source), both of which I'm trying to avoid for this script.

As an alternative, I'm considering baking in something that will detect if fio is installed and available on the user's system and use that in place of dd/ioping, if detected. This will get around both requirements (elevated privileges and installed prereqs), but can still use a more accurate disk performance test if the user has fio installed.

from yet-another-bench-script.

I'm considering baking in something that will detect if fio is installed and available on the user's system and use that in place of dd/ioping, if detected

This is a good idea. I wouldn't even use dd as a fallback; I'd just print out an error message like "cannot do disk I/O speed test unless fio is installed"

What about building a statically-linked version of fio and downloading it in the script? That might work too.

from yet-another-bench-script.

What about building a statically-linked version of fio and downloading it in the script? That might work too.

If I can get it to compile and work on other machines, I'll definitely do that since it's the same approach I'm taking with ioping. The couple attempts I made at this resulted in the compiled binary not working correctly on other machines (some it did, some it didn't). I'll take another stab at it soon and see if I can get it to compile and to run on other machines without any issues. I'll likely have to compile a 32-bit and a 64-bit version, but that's fine.

from yet-another-bench-script.

fio implemented within 2c74b1a

from yet-another-bench-script.

Thank you @masonr! This is great :D

from yet-another-bench-script.