Comments on: Linux Precision http://blog.sidneidasilva.com/2008/08/22/linux-precision/ Sat, 31 Jul 2010 20:09:21 +0000 hourly 1 http://wordpress.com/ By: Buchan http://blog.sidneidasilva.com/2008/08/22/linux-precision/#comment-18557 Buchan Fri, 29 Aug 2008 13:47:42 +0000 http://omsg.wordpress.com/?p=264#comment-18557 So, critics haven't figured out that Avail != Size - Used ? Avail == Size - Used - Reserved Maybe instead of complaining, users should ask for the reserved space to be shown in the df output. Or, as end users, using a graphical tool (such as the properties dialog in Konqueror/Dolphin, which bases its values on the non-reserved ). For example, on one of my machines: $ df -P /home Filesystem 1024-blocks Used Available Capacity Mounted on /dev/mapper/VGsystem-home 35079261 33545678 1533583 96% /home $ df /home Filesystem 1K-blocks Used Available Use% Mounted on /dev/mapper/VGsystem-home 35079261 33545598 1533663 96% /home But, Konqueror shows: Free disk space: 1.5GB out of 33.5GB People complain that linux is like 'DOS', but then complain when df gives them an answer they can't understand, while all the GUI tools give them the answer in the way they can understand it. I give up. So, critics haven’t figured out that Avail != Size – Used ?

Avail == Size – Used – Reserved

Maybe instead of complaining, users should ask for the reserved space to be shown in the df output. Or, as end users, using a graphical tool (such as the properties dialog in Konqueror/Dolphin, which bases its values on the non-reserved ). For example, on one of my machines:

$ df -P /home
Filesystem 1024-blocks Used Available Capacity Mounted on
/dev/mapper/VGsystem-home 35079261 33545678 1533583 96% /home
$ df /home
Filesystem 1K-blocks Used Available Use% Mounted on
/dev/mapper/VGsystem-home
35079261 33545598 1533663 96% /home

But, Konqueror shows:
Free disk space: 1.5GB out of 33.5GB

People complain that linux is like ‘DOS’, but then complain when df gives them an answer they can’t understand, while all the GUI tools give them the answer in the way they can understand it. I give up.

]]> By: Owen http://blog.sidneidasilva.com/2008/08/22/linux-precision/#comment-18556 Owen Fri, 29 Aug 2008 12:54:21 +0000 http://omsg.wordpress.com/?p=264#comment-18556 Why not use du -hs / if you want to know the size of everything on your drive. Or better yet, use xdiskusage. I thought some of the "lost" space in the df output was from inodes. Why not use du -hs / if you want to know the size of everything on your drive. Or better yet, use xdiskusage. I thought some of the “lost” space in the df output was from inodes.

]]> By: ChirpyChirpyCheapCheap http://blog.sidneidasilva.com/2008/08/22/linux-precision/#comment-18555 ChirpyChirpyCheapCheap Wed, 27 Aug 2008 10:59:28 +0000 http://omsg.wordpress.com/?p=264#comment-18555 man df -h, --human-readable print sizes in human readable format (e.g., 1K 234M 2G) Examples show it's rounding. Here're the alternatives: 18:58 [summer@numbat ~]$ df -h Filesystem Size Used Avail Use% Mounted on /dev/mapper/VolGroup00-LogVol00 35G 32G 1.1G 97% / /dev/hda1 99M 61M 34M 65% /boot tmpfs 1.5G 0 1.5G 0% /dev/shm 18:58 [summer@numbat ~]$ df Filesystem 1K-blocks Used Available Use% Mounted on /dev/mapper/VolGroup00-LogVol00 35772016 32862740 1062800 97% / /dev/hda1 101086 62001 33866 65% /boot tmpfs 1481572 0 1481572 0% /dev/shm 18:58 [summer@numbat ~]$ Which is easier to read? man df
-h, –human-readable
print sizes in human readable format (e.g., 1K 234M 2G)
Examples show it’s rounding. Here’re the alternatives:
18:58 [summer@numbat ~]$ df -h
Filesystem Size Used Avail Use% Mounted on
/dev/mapper/VolGroup00-LogVol00
35G 32G 1.1G 97% /
/dev/hda1 99M 61M 34M 65% /boot
tmpfs 1.5G 0 1.5G 0% /dev/shm
18:58 [summer@numbat ~]$ df
Filesystem 1K-blocks Used Available Use% Mounted on
/dev/mapper/VolGroup00-LogVol00
35772016 32862740 1062800 97% /
/dev/hda1 101086 62001 33866 65% /boot
tmpfs 1481572 0 1481572 0% /dev/shm
18:58 [summer@numbat ~]$
Which is easier to read?

]]> By: Evil Closet Monkey http://blog.sidneidasilva.com/2008/08/22/linux-precision/#comment-18554 Evil Closet Monkey Tue, 26 Aug 2008 14:35:21 +0000 http://omsg.wordpress.com/?p=264#comment-18554 Some appear to be forgetting what Sid is looking at is "human readable" output. Look up through the comments above and you see all the excuses and reasons why the output is correct... when it really isn't! Waylon hits it on the head. 'df' might be doing what it is supposed to, but the output is useless to a regular user. You shouldn't have to understand what is happening under the hood to understand the output. @Joe: Remember that is is all about the user, user, *user*! If something doesn't make sense, it's wrong. Some appear to be forgetting what Sid is looking at is “human readable” output. Look up through the comments above and you see all the excuses and reasons why the output is correct… when it really isn’t!

Waylon hits it on the head. ‘df’ might be doing what it is supposed to, but the output is useless to a regular user. You shouldn’t have to understand what is happening under the hood to understand the output.

@Joe: Remember that is is all about the user, user, *user*! If something doesn’t make sense, it’s wrong.

]]> By: Joe http://blog.sidneidasilva.com/2008/08/22/linux-precision/#comment-18552 Joe Tue, 26 Aug 2008 06:56:54 +0000 http://omsg.wordpress.com/?p=264#comment-18552 If a person doesn't care about df, it doesn't matter if the number make sense to them or not. df is merely reporting the values returned by the file system driver. The column headings even make sense, if you actually read them. The "size" is the size of the file system, the "used" is how much is used and the "avail" is how much is available for additional writing in normal operation. No where does it say that these need to add up to the same amounts, they are mostly completely unrelated values. A full disk is an exceptional condition, and the system is set up to avoid serious problems during an exceptional condition. If a person doesn’t care about df, it doesn’t matter if the number make sense to them or not.

df is merely reporting the values returned by the file system driver. The column headings even make sense, if you actually read them. The “size” is the size of the file system, the “used” is how much is used and the “avail” is how much is available for additional writing in normal operation. No where does it say that these need to add up to the same amounts, they are mostly completely unrelated values. A full disk is an exceptional condition, and the system is set up to avoid serious problems during an exceptional condition.

]]> By: sidnei http://blog.sidneidasilva.com/2008/08/22/linux-precision/#comment-18551 sidnei Tue, 26 Aug 2008 00:34:47 +0000 http://omsg.wordpress.com/?p=264#comment-18551 @Waylon: You're absolutely right. Not even 'man df' gives you a clue about what's going on. @Waylon: You’re absolutely right. Not even ‘man df’ gives you a clue about what’s going on.

]]> By: Waylon Kenning http://blog.sidneidasilva.com/2008/08/22/linux-precision/#comment-18550 Waylon Kenning Tue, 26 Aug 2008 00:27:56 +0000 http://omsg.wordpress.com/?p=264#comment-18550 I think the point is for a person who doesn't care about df, those numbers don't even make sense. It doesn't even mention rounding. This is a perfect example of the program being 'technically correct', but not being useful. Well done. I think the point is for a person who doesn’t care about df, those numbers don’t even make sense. It doesn’t even mention rounding.

This is a perfect example of the program being ‘technically correct’, but not being useful. Well done.

]]> By: Matt Curtin http://blog.sidneidasilva.com/2008/08/22/linux-precision/#comment-18545 Matt Curtin Sat, 23 Aug 2008 15:14:41 +0000 http://omsg.wordpress.com/?p=264#comment-18545 What is happening here goes back far before Linux. A search in the literature will show references to “minfree” (which is the amount “reserved for root”). BSD's tunefs(8) man page describes this in some more detail. Secondly, it is worth pointing out that you're complaining not about a lack of “precision” but of accuracy. They aren't the same thing. And once you know how to read the df(1) output, of course, you will see that there is no accuracy problem at all. What is happening here goes back far before Linux. A search in the literature will show references to “minfree” (which is the amount “reserved for root”). BSD’s tunefs(8) man page describes this in some more detail.

Secondly, it is worth pointing out that you’re complaining not about a lack of “precision” but of accuracy. They aren’t the same thing. And once you know how to read the df(1) output, of course, you will see that there is no accuracy problem at all.

]]> By: Dorneles Treméa http://blog.sidneidasilva.com/2008/08/22/linux-precision/#comment-18544 Dorneles Treméa Sat, 23 Aug 2008 04:59:00 +0000 http://omsg.wordpress.com/?p=264#comment-18544 Hey Sid, Alexandre is right... by default, when creating a new file system with mkfs.ext3, it will reserve 5% to root. This percentage is taken into account even when running df with root privileges. To recover your extra 1GB back, just run: sudo tune2fs -m 0 /dev/sda2 BTW, it recovered 25GB on my half terabyte HD... :-P Take care. -- deo Hey Sid,

Alexandre is right… by default, when creating a new file system with mkfs.ext3, it will reserve 5% to root.

This percentage is taken into account even when running df with root privileges.

To recover your extra 1GB back, just run:

sudo tune2fs -m 0 /dev/sda2

BTW, it recovered 25GB on my half terabyte HD… :-P

Take care.

– deo

]]> By: sidnei http://blog.sidneidasilva.com/2008/08/22/linux-precision/#comment-18543 sidnei Fri, 22 Aug 2008 22:57:16 +0000 http://omsg.wordpress.com/?p=264#comment-18543 I did, same results. But that might be a good clue as to what's going on. I did, same results. But that might be a good clue as to what’s going on.

]]>