Please help me understand this article

[cjxzjcxzjz]

How to Manage Linux File System on Ubuntu Server – LinuxWays

linuxways.net

Bottom section:

it will show you request size as avgrq-sz which shows how your disk is performing against requests.

I don't see "avgrq-sz" in his screenshot and also not on my own system, what am I missing? Should I interpret this in a particular way?

If your disk’s avgrq-sz is higher than 100.0 then it means it is performing optimally. Otherwise, you will need to improve it.

Could it be that he means higher than 1.0 instead here?

 

[NorthWest]

what am I missing?

I think you are missing the information, or part of the information, that is in the manpage of iostat.

 

[KGIII]

I played around with this last night, not long before going to bed. I couldn't get "avgrq-sz" to show no matter what I did. If should appear if you use the -x flag, from what I read. It never appeared in any of my iostat results.

 

[NorthWest]

This is the quote from the iostat man page that addresses the options:

areq-sz
The average size (in kilobytes) of the I/O requests that were issued to the device.
Note: In previous versions, this field was known as avgrq-sz and was expressed in sectors.

"avgrq-sz' has been replaced by "areq-sz" the latter expressing the I/O requests in a different measure. So the measure as an output of iostat is evidently not gone, rather, changed.

But, as you found, KGill, it doesn't appear in the output of "iostat -x", or any other option I tried.

I've had a look at the source code over at https://github.com/sysstat/sysstat/blob/master/iostat.c, and "areq-sz" is part of an "if ... else" statement in which the condition must simply have not been fulfilled to print it in the isostat command that we have, so the code goes on to print "rareq-sz". I'll leave the description of the condition to print "areq-sz" to the interested inquirer. That code was amended 4 days ago so is not the code used in the suite of sysstat commands which users are using currently supplied by their distributions at the moment. It takes time for these things to move through to the distributions. So, it's possible, and this is conjecture, that the absence of the option's output has been attended to. I haven't compiled the code to check that though.

 

[cjxzjcxzjz]

Thank you all for the replies and info. NorthWest even checking source code to find a possible clue which is very much appreciated. Cheers all.