Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 09 Apr 2015 16:39:00 -0500
From:      Rick Romero <rick@havokmon.com>
To:        freebsd-fs <freebsd-fs@freebsd.org>
Subject:   ZFS 10.1 send single snapshot - space 'used' irregularity
Message-ID:  <20150409163900.Horde.ZLVwr91i2UaonmJT1bC-Pw1@www.vfemail.net>

next in thread | raw e-mail | index | archive | help
I have 3 servers, A, B, C.  I'm building C to replace A, and replicating
the data to C from backup B.  A is offsite in relation to B and C.
All servers are FreeBSD 10.1, except A - which is 9.2.

I'm confused on disk usage. Not so much a GB here or there, but 250GB is
'unaccounted for' on C.  C and A should be a pretty close match.

A - looks correct

sysvolssd2/home  used                 
495G                   -
sysvolssd2/home  usedbysnapshots      
37.9G                  -
sysvolssd2/home  usedbydataset        
456G                   -
sysvolssd2/home  usedbychildren       
669M                   -
sysvolssd2/home  usedbyrefreservation 
0                      -
sysvolssd2/home  logicalused          
585G                   -

NAME         SIZE     ALLOC   FREE    CAP   
DEDUP    HEALTH  ALTROOT
sysvolssd2  1.39T   744G   680G       52%     
1.00x     ONLINE  -

B - looks correct (backup of A, holds more snapshots and other crap than A)
sysvol/primessd_home  used                 
777G                   -
sysvol/primessd_home  usedbysnapshots      
240G                   -
sysvol/primessd_home  usedbydataset        
537G                   -
sysvol/primessd_home  usedbychildren       
0                      -
sysvol/primessd_home  usedbyrefreservation 
0                      -
sysvol/primessd_home  logicalused          
754G                   -

NAME         SIZE  ALLOC   FREE   FRAG  EXPANDSZ    CAP 
DEDUP  HEALTH  ALTROOT
sysvol          4.53T  2.43T  2.10T    20%        
-                 53%  1.00x  ONLINE  - 

C - missing what appears to be the multiple snapshot data.  Only the
latest snapshot was sent, not the entire dataset.  So 531GB is close
enough to the 537G of B's dataset.
sysvol_enc/home  used                 
758G                   -
sysvol_enc/home  usedbysnapshots      
3.00M                  -
sysvol_enc/home  usedbydataset        
752G                   -
sysvol_enc/home  usedbychildren       
5.84G                  -
sysvol_enc/home  usedbyrefreservation 
0                      -
sysvol_enc/home  logicalused          
531G                   -
NAME         SIZE  ALLOC   FREE   FRAG  EXPANDSZ    CAP 
DEDUP  HEALTH  ALTROOT
sysvol_enc  1.39T  1.12T   277G    49%        
-                   80%  1.00x  ONLINE  -

C is geli encrypted and B is not.

Unfortunately when I check another server that's geli encrypted, it looks
fine:

E -
nlsysvol/home  used                 
13.8G                  -
nlsysvol/home  usedbysnapshots      
5.58G                  -
nlsysvol/home  usedbydataset        
7.78G                  -
nlsysvol/home  usedbychildren       
483M                   -
nlsysvol/home  usedbyrefreservation 
0                      -
nlsysvol/home  logicalused          
12.0G                  -
NAME       SIZE  ALLOC   FREE   FRAG  EXPANDSZ    CAP 
DEDUP  HEALTH  ALTROOT
nlsysvol   115G  42.8G    72.2G      -        
-                    37%  1.00x    ONLINE  -

So the difference shouldn't be related to the encryption.  It's almost as
if the send from B to C included all the incremental snapshots, but didn't
actually account for them.  Am I reading this wrong, or is something else
not right ?
Should I delete that dataset, re-send the entire original dataset, then
delete the incremental snapshots?

It makes me a little concerned that deleting a snapshot might delete the
data which was written at that time, even though it was not deleted in
followup snapshots...
And I assume FRAG is fragmentation.  50% is a bit strange for a brand new
receive, isn't it?

help.  :)

Rick



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20150409163900.Horde.ZLVwr91i2UaonmJT1bC-Pw1>