AnsweredAssumed Answered

HNAS mount point with very different outputs of "df" and "du"

Question asked by Joao Serra on Sep 30, 2016

As seen on the HNAS:

 

hnas-1[evs-2]:$ filesystem-list projpwx_prod

filesystem-list: executing on cluster node 1, though the EVS in context (2) is currently on cluster node 2

Instance name      Dev   On span            State  EVS  Cap/GiB  Confined Flag

-----------------  ----  -----------------  -----  ---  -------  -------- ----

projpwx_prod       1031  HNAS_POOL_1        Mount   2       126      110

 

Mounted on server 10.191.37.227:

10.191.37.227:/projpwx_prod on /projpwx type nfs (rw,addr=10.191.37.227)

 

The DF command reports the right size:

 

[root@supwc101 /]# df -h /projpwx

Filesystem            Size  Used Avail Use% Mounted on

10.191.37.227:/projpwx_prod

                      126G   76G   51G  60% /projpwx

 

[root@supwc101 /]# df  /projpwx

Filesystem           1K-blocks      Used Available Use% Mounted on

10.191.37.227:/projpwx_prod

                     132055040  78693696  53361344  60% /projpwx

 

Yet, running DU reports a much lower used capacity.

                    

[root@supwc101 /]# du -hs projpwx

35G     projpwx

[root@supwc101 /]# du -s projpwx

35690424        projpwx

 

I suspect this difference is due to fs metada (there are over 19 milion inodes used) and/or because "df" rounds up its accounting to the 4K blocks and "du" sums the real files size.

Please let me know if I'm right in any aspect.

Outcomes