Home / server / Questions / 443151
In Process
Guillaume
Asked: 2012-10-29 07:23:56 +0800 CST

Too much free space on FreeNAS - ZFS

  • 772

I have a FreeNAS server with 3 x 2 To disks in raidz1. I would expect to have about 4 To of space available. When I run zpool list I get:

[root@freenas] ~# zpool list
NAME          SIZE   USED  AVAIL    CAP  HEALTH  ALTROOT
main_volume  5.44T  3.95T  1.49T    72%  ONLINE  /mnt

I was expecting a size of 4 To.

Also, used space as reported by zpool list does not match what's reported by du:

[root@freenas] ~# du -sh /mnt/main_volume/
2.6T    /mnt/main_volume/

There are quite a few things that I dont yet completely understand about ZFS. But at the moment I am mostly worried that I misconfigured my system and that I dont have any storage redundancy. How can I make sure I did not do an horrible mistake ...

For the sake of completeness, here is the output of zpool status:

[root@freenas] ~# zpool status
  pool: main_volume
 state: ONLINE
 scrub: none requested
config:

    NAME                                            STATE     READ WRITE CKSUM
    main_volume                                     ONLINE       0     0     0
      raidz1                                        ONLINE       0     0     0
        gptid/d8584e45-5b8a-11d9-b9ea-5404a6630115  ONLINE       0     0     0
        gptid/d8f7df30-5b8a-11d9-b9ea-5404a6630115  ONLINE       0     0     0
        gptid/d9877cc3-5b8a-11d9-b9ea-5404a6630115  ONLINE       0     0     0

errors: No known data errors
zfs

2 Answers

  1. zpool list shows the RAW size of your disks (e.g. without RAID).

    That is why it appears larger than what you have available in zfs list.

    The setup you've listed shows a RAIDZ1 zpool. That will have the usable space of n-1 disks.

    Here's a good primer to understanding ZFS.

    • 5

  2. Looks OK to me.
    In the zpool list output the USED column should be read as USABLE.
    And AVAIL is free within that usable space.
    That fits nicely with your DU output. (Slight discrepancy is due to roundoff and not counting filesystem and ZFS meta-data.)

    • 2