Hi all
I just started testing iSCSI volumes from my Solaris 11.3 server to my Windows 10 workstation. It's working great, and was super easy to setup.
But I am seeing one very confusing thing: zfs list always reports far more data used in the ZFS volume than I have actually created.
I created a 512GB ZFS volume with no reservation (zfs create -sV), created a LUN and target, and then mounted it on Windows and formatted it with NTFS.
In Windows I then copied 322GB of data onto the new drive. That's all completely fine. But zfs list shows the space used in the volume as over double what has actually been used?
This is what I see with zfs list and zfs get:
Everything is working fine, but I just don't understand the values the zfs command is showing? The used figure is 2.265 times higher than the actual data used as reported in Windows - a very strange ratio that I can't begin to figure out.
Here's another example, this time using a standard volume, and writing to it from Solaris:
In this example I created a 128G volume, turned compression off, created an iSCSI LUN and target and then accessed that target via the local initiator. I created a pool on the iSCSI disk and loaded a total of 107GB onto it.
But again, zfs shows crazy figures - including a USED value of 248GB for a 128GB volume! This time the ratio of USED compared to the actual data is 2.303 - very similar to the previous 2.65, but not identical.
What is up with this? Clearly there's some logic to what this figure is, but I can't begin to understand it. Is it related to iSCSI, or is it a general issue with ZFS volumes? I've not really used volumes before, and never closely checked the ones I did create.
It's a little annoying to not see the correct usage figures in the output of zfs. It will make it harder to work out actual space used.
Thanks in advance
I just started testing iSCSI volumes from my Solaris 11.3 server to my Windows 10 workstation. It's working great, and was super easy to setup.
But I am seeing one very confusing thing: zfs list always reports far more data used in the ZFS volume than I have actually created.
I created a 512GB ZFS volume with no reservation (zfs create -sV), created a LUN and target, and then mounted it on Windows and formatted it with NTFS.
In Windows I then copied 322GB of data onto the new drive. That's all completely fine. But zfs list shows the space used in the volume as over double what has actually been used?
This is what I see with zfs list and zfs get:
Code:
root@magrathea:~# zfs list -rt all data/volumes
NAME USED AVAIL REFER MOUNTPOINT
data/volumes 736G 30.7T 347K /data/volumes
data/volumes/iscsi 736G 30.7T 329K /data/volumes/iscsi
data/volumes/iscsi/test-512g 736G 30.7T 736G -
root@magrathea:~# zfs get volsize,used,compress,compressratio data/volumes/iscsi/test-512g
NAME PROPERTY VALUE SOURCE
data/volumes/iscsi/test-512g volsize 512G local
data/volumes/iscsi/test-512g used 736G -
data/volumes/iscsi/test-512g compression lz4 inherited from data
data/volumes/iscsi/test-512g compressratio 1.08x -
root@magrathea:~# zfs get -p used data/volumes/iscsi/test-512g
NAME PROPERTY VALUE SOURCE
data/volumes/iscsi/test-512g used 789776179920 -
Here's another example, this time using a standard volume, and writing to it from Solaris:
Code:
root@magrathea:~# zfs create -V 128G data/volumes/iscsi/test-128g
root@magrathea:~# zfs set compress=off data/volumes/iscsi/test-128g
root@magrathea:~# zfs list data/volumes/iscsi/test-128g
NAME USED AVAIL REFER MOUNTPOINT
data/volumes/iscsi/test-128g 132G 30.7T 165K -
root@magrathea:~# stmfadm create-lu /dev/zvol/rdsk/data/volumes/iscsi/test-128g
Logical unit created: 600144F069834D00000058EADB1B0003
root@magrathea:~# stmfadm add-view 600144F069834D00000058EADB1B0003
root@magrathea:~# itadm create-target
Target iqn.1986-03.com.sun:02:5527fe9e-d137-40b8-9c65-f51550002283 successfully created
root@magrathea:~# iscsiadm add discovery-address 192.168.0.5
root@magrathea:~# iscsiadm modify discovery --sendtargets enable
root@magrathea:~# zpool create iscsi-test-128g c0t600144F069834D00000058EADB1B0003d0
root@magrathea:~# zpool list iscsi-test-128g
NAME SIZE ALLOC FREE CAP DEDUP HEALTH ALTROOT
iscsi-test-128g 127G 98.5K 127G 0% 1.00x ONLINE -
root@magrathea:~# gdd if=/dev/zero of=/iscsi-test-128g/file bs=128k count=80000
80000+0 records in
80000+0 records out
10485760000 bytes (10 GB) copied, 15.1727 s, 691 MB/s
root@magrathea:~# gdd if=/dev/zero of=/iscsi-test-128g/file2 bs=128k count=800000
800000+0 records in
800000+0 records out
104857600000 bytes (105 GB) copied, 233.563 s, 449 MB/s
root@magrathea:~# zfs list iscsi-test-128g data/volumes/iscsi/test-128g
NAME USED AVAIL REFER MOUNTPOINT
data/volumes/iscsi/test-128g 248G 30.4T 248G -
iscsi-test-128g 107G 17.6G 107G /iscsi-test-128g
root@magrathea:~# zfs get volsize,refer,refreservation,usedbydataset,usedbyrefreservation data/volumes/iscsi/test-128g
NAME PROPERTY VALUE SOURCE
data/volumes/iscsi/test-128g volsize 128G local
data/volumes/iscsi/test-128g referenced 248G -
data/volumes/iscsi/test-128g refreservation 132G default
data/volumes/iscsi/test-128g usedbydataset 248G -
data/volumes/iscsi/test-128g usedbyrefreservation 0 -
data/volumes/iscsi/test-128g compressratio 1.00x -
In this example I created a 128G volume, turned compression off, created an iSCSI LUN and target and then accessed that target via the local initiator. I created a pool on the iSCSI disk and loaded a total of 107GB onto it.
But again, zfs shows crazy figures - including a USED value of 248GB for a 128GB volume! This time the ratio of USED compared to the actual data is 2.303 - very similar to the previous 2.65, but not identical.
What is up with this? Clearly there's some logic to what this figure is, but I can't begin to understand it. Is it related to iSCSI, or is it a general issue with ZFS volumes? I've not really used volumes before, and never closely checked the ones I did create.
It's a little annoying to not see the correct usage figures in the output of zfs. It will make it harder to work out actual space used.
Thanks in advance