This is normal behavior for XIV LUNs
In a thick pool space is dynamically allocated as needed, similar to a thin LUN. However, since there is enough space in the pool for all of the LUNs to be fully allocated you will never run into a situation where you run out of space in the pool. The advatages to running reclamation against a thick pool are as follows;
1) when there is disk/module failure, the data which needs to be rebuilt will reduce because of space reclamation
2) it can increase the spare capacity for rebuild
However, IBM support has confirmed that LUNs which are mirrored on the array or associated with an array-based snapshot do not support thin reclamation. If a snapshot or mirror is created on the array, then the customer should run "vxdctl enable" to rescan the current LUN attributes. The "thinrclm" flag should then disappear on the LUN in question. We detect whether or not the LUN supports reclamation by checking byte 5 of page 0xc0 of the SCSI inquiry.
0x00 is thick, while 0x01 is thin.
For example, both of these LUNs are part of the same pool, yet they have different values for byte 5 of page 0xc0;
# ./vxscsiinq -d -e 1 -p 192 /dev/rdsk/c3t500173804ECD0150d29s2
Inquiry for /dev/rdsk/c3t500173804ECD0150d29s2, evpd 0x1, page code 0xc0
/dev/rdsk/c3t500173804ECD0150d29s2: Raw data size 28
Bytes 0 - 9 0x00 0xc0 0x00 0x18 0x00 0x01 0x00 0x02 0x00 0x10
Bytes 10 - 19 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x80 0x00 0x00
Bytes 20 - 27 0x00 0x00 0x00 0x00 0x10 0x00 0x00 0x00
Vs
# ./vxscsiinq -d -e 1 -p 192 /dev/rdsk/c3t500173804ECD0140d3s2
Inquiry for /dev/rdsk/c3t500173804ECD0140d3s2, evpd 0x1, page code 0xc0
/dev/rdsk/c3t500173804ECD0140d3s2: Raw data size 28
Bytes 0 - 9 0x00 0xc0 0x00 0x18 0x00 0x00 0x00 0x02 0x00 0x10
Bytes 10 - 19 0x00 0x00 0x00 0x00 0x00 0x04 0x00 0x50 0x00 0x00
Bytes 20 - 27 0x00 0x00 0x00 0x00 0x10 0x00 0x00 0x00
The first example shows a LUN that supports reclamation, while the second example shows a LUN which does not. The difference is that the second LUN has a snapshot associated with it on the array. Therefore the bit changes to 0. Next time "vxdctl enable" is run the flags will be updated to reflect this.
Here is an example of how various types of XIV LUNs are expected to appear in VxVM
# vxdisk -e list
DEVICE TYPE DISK GROUP STATUS OS_NATIVE_NAME ATTR
disk_0 auto:none - - online invalid c0t0d0s2 -
disk_1 auto:none - - online invalid c0t1d0s2 -
xiv0_020a auto:cdsdisk xiv0_020a thindg online thinrclm c3t50017380652B0142d1s2 tprclm RAID_10 <---- This LUN is part of a thin pool with no mirrors or snapshots
xiv0_0207 auto:cdsdisk - - online c3t50017380652B0150d4s2 std RAID_10 <---- This LUN is part of the same thin pool, and it has an associated snapshot on the array
xiv0_0208 auto:cdsdisk xiv0_0208 thickdg online thinrclm c3t50017380652B0142d3s2 tprclm RAID_10 <---- This LUN is part of a thick pool with no mirrors or snapshots
xiv0_0209 auto:cdsdisk - - online c3t50017380652B0142d2s2 std RAID_10 <---- This LUN is part of a thick pool, and it has an associated snapshot on the array.
xiv0_524 auto:cdsdisk - - online udid_mismatch c3t50017380652B0142d5s2 std snapshot optimized_snapshot <---- This is an array based snapshot of LUN xiv0_0209
Note that LUNs with associated snapshots do not have any flag to indicate this. They simply show up without the thinrclm flag. If the snapshot itself is presented from the array, then it will show up with a "snapshot" flag.
Applies To
IBM XIV array supporting thin reclamation