Where'd my data go? snapshots

[G Brown]

Trying to understand snapshot backups, send, recv, etc.

I made a recursive snapshot rz2/data which has rz2/data/archive and rz2/data/gb, and sent this to the "offsite" zpool. Fine.

df says:

Code:

Filesystem                    Size    Used   Avail Capacity  Mounted on
rz2/data                      4.6T    232k    4.6T    0%    /mnt/rz2/data
rz2/data/archive              5.6T    991G    4.6T    17%    /mnt/rz2/data/archive
rz2/data/gb                  5.1T    474G    4.6T    9%    /mnt/rz2/data/gb
offsite                      2.1T    160k    2.1T    0%    /mnt/offsite
offsite/data                  2.1T    160k    2.1T    0%    /mnt/offsite/data
offsite/data/archive          3.1T    991G    2.1T    31%    /mnt/offsite/data/archive
offsite/data/gb              2.6T    475G    2.1T    18%    /mnt/offsite/data/gb

So storage space is being used.
There is only one snapshot, so I shouldn't be too confused.

Code:

zfs list -t snapshot
NAME                                  USED  AVAIL  REFER  MOUNTPOINT
offsite/data@manual-20140302            8K      -  160K  -
offsite/data/archive@manual-20140302    8K      -  991G  -
offsite/data/gb@manual-20140302          8K      -  475G  -
offsite/test@1                          8K      -  152K  -
rz2/data@manual-20140302              163K      -  232K  -
rz2/data/archive@manual-20140302      16.4M      -  991G  -
rz2/data/gb@manual-20140302          3.32G      -  477G  -
rz2/test@1                                0      -  221K  -

But when I try to list the files, they are absent:

Code:

>>>freenas: root@/mnt/offsite$: ll data/*
data/archive:
total 34
drwxr-xr-x  2 root  wheel  - 2 Mar  1 16:07 ./
drwxr-xr-x  4 root  wheel  - 4 Mar  1 16:07 ../
 
data/gb:
total 34
drwxr-xr-x  2 root  wheel  - 2 Mar  1 16:07 ./
drwxr-xr-x  4 root  wheel  - 4 Mar  1 16:07 ../

I could re-send the data, but it looks like it is there, just not visible.

I am curious as to where it went.

Thanks for any insights.

 

[G Brown]

Well I made new snapshots, re-inited the offsite pool, sent them to the new "offsite", and scripted it all so bad things won't happen.

Checksums, ECC ram, etc. but I think a mis-type or software bug will contribute more to mean-time-to-data-loss than any of the hardware issues. Its the UI.

 

[cyberjock]

For starters, df is a mess with ZFS. It doesn't understand things such as compression, deduplication, snapshots, etc. Pretty much anyone using df or du to monitor ZFS in any kind of diagnostic manner is wasting their time. ZFS has it's own commands for monitoring disk space. zpool list, zfs list, etc.