[lug] df discrepancies between used, available and total size

[karl horlen]

i just reinstalled a server from scratch using centos 6.x and two brand new 1T drives.


i basically put LVM on top of a software 1 raid mirror.

immediately after install, i was surprised df -ha reported what appears to be some major discrepancies between how much was being used vs how much was available on some of my LVs.  the available amoutn on some of the LVs is a lot less than the size minus the used by a considerable margin.

example:

size:769G
used:197M (yeah that's an M)
avail: 730G

that's like a 39G discrepancy (wasted space) on my /home lv partition.  wtf?

two other LVs arent as bad but they're a lot smaller so the hist doesn't look as bad

size:12G
used:1.7G
avail: 9.4G

i'm losing almost 1 G there

size:15G
used:241M
avail:14G

i'm losing 0.75 G there

the first lv is obvious though


is this normal?.  i can't recall seeing such wide gaps on servers i've installed / managed over the years.  was i not paying close enough attention?  am i missing something?   is there space set aside / reserved when the partition is relatively empty?  is my block size somehow screwed up?  it can't be bad sectors b/c that would be reported in the overall size?

just wondering if i should be checking for something or if someone can enlighten me about that 39G missing space

thanks