ONTAP Discussions
ONTAP Discussions
Hi,
we buy a new FAS2720 with 4 SSD and 8 HDD and 3 DS212C with 12 HDD each, for a total of 4 SSD 960GB and 44 HDD 4TB.
Now I'm configuring Flash Pool, like the old FAS2554 (4 SSD + 20 HDD) that will be migrated on the new FAS.
So I use the same command use some years ago on the 2554:
storage aggregate modify -aggregate cluster4_01_NL_SAS_1 -hybrid-enabled true
storage pool create -storage-pool StoragePool_c4_Cache1 -disk-list 1.0.0,1.0.10,1.0.1
now
storage aggregate show-spare-disks -disk-type SSD
show one disk as spare on node02 (the same as 2554).
> storage aggregate show-spare-disks -disk-type SSD
Original Owner: cluster4-02
Pool0
Spare Pool
Usable Physical
Disk Type Class RPM Checksum Size Size Status
---------------- ------ ----------- ------ -------------- -------- -------- --------
1.0.11 SSD solid-state - block 894.0GB 894.3GB zeroed
But when I try to add flash-pool to aggregate I have this error:
storage aggregate add cluster4_01_NL_SAS_1 -storage-pool StoragePool_c4_Cache1 -allocation-units 4 -raidtype raid4
Error: command failed: storage pool "StoragePool_c4_Cache1" does not have enough spare allocation units. Use "storage pool show-available-capacity" to get the spare allocation units assigned to node cluster4-01.
storage pool show-available-capacity
Storage SyncMirror Allocation Unit Total
Node Storage Pool Type Pool Unit size Count Usable Size
---------- --------------- ------- ---------- ---------- ----- -----------
cluster4-01
StoragePool_c4_Cache1
SSD Pool0 670.5GB 2 1.31TB
cluster4-02
StoragePool_c4_Cache1
SSD Pool0 670.5GB 2 1.31TB
2 entries were displayed.
So, what is changed on FAS2720 and/or ONTAP 9.9.1? I remember that in the past an SSD spare can be shared from node.
Now I'm in difficulty if I cannot create my flash pool.
Thanks
Solved! See The Solution
By default, two allocation units are auto-assigned to each node. That is why when you try to specify 4 units, it is failing. It would probably work with 2 units
If you >>really<< want all four units to be part of the same aggregate, you will need to reassign the other two units:
By default, two allocation units are auto-assigned to each node. That is why when you try to specify 4 units, it is failing. It would probably work with 2 units
If you >>really<< want all four units to be part of the same aggregate, you will need to reassign the other two units:
Ok, probably understand what is wrong, on my documentation I write "-allocation-units 4" but was "2", because I need to add flash pool to 2 aggregate and if reassign a total of 4 AU to only one node is impossible to do this.
Here is the status of flash-pool on current FAS 2554 that I want replicated on FAS 2720:
cluster2::> storage pool show-disks -storage-pool ClusterCache
Storage Pool Name: ClusterCache
Disk
Disk Type Usable Size Total Size
-------- ------- ----------- -----------
1.0.0 SSD 372.4GB 372.6GB
1.0.1 SSD 372.4GB 372.6GB
1.0.2 SSD 372.4GB 372.6GB
3 entries were displayed.
cluster2::> storage pool show-disks -storage-pool ClusterCache -instance
Name of Storage Pool: ClusterCache
Name of the disk: 1.0.0
Disk Type: SSD
Disk Usable Size: 372.4GB
Total Size: 372.6GB
List of Nodes: cluster2-node01, cluster2-node02
Name of Storage Pool: ClusterCache
Name of the disk: 1.0.1
Disk Type: SSD
Disk Usable Size: 372.4GB
Total Size: 372.6GB
List of Nodes: cluster2-node01, cluster2-node02
Name of Storage Pool: ClusterCache
Name of the disk: 1.0.2
Disk Type: SSD
Disk Usable Size: 372.4GB
Total Size: 372.6GB
List of Nodes: cluster2-node01, cluster2-node02
3 entries were displayed.
cluster2::> storage pool show-disks -storage-pool ClusterCache
Storage Pool Name: ClusterCache
Disk
Disk Type Usable Size Total Size
-------- ------- ----------- -----------
1.0.0 SSD 372.4GB 372.6GB
1.0.1 SSD 372.4GB 372.6GB
1.0.2 SSD 372.4GB 372.6GB
3 entries were displayed.
cluster2::> storage aggregate show-spare-disks -disk-type SSD
Original Owner: cluster2-node01
Pool0
Spare Pool
Usable Physical
Disk Type Class RPM Checksum Size Size Status
---------------- ------ ----------- ------ -------------- -------- -------- --------
1.0.3 SSD solid-state - block 372.4GB 372.6GB zeroed
cluster2::> storage pool show -storage-pool ClusterCache -instance
Storage Pool Name: ClusterCache
UUID of Storage Pool: 3a072807-eec0-11e6-bd65-00a098c6ed93
Nodes Sharing the Storage Pool: cluster2-node02, cluster2-node01
Number of Disks in Storage Pool: 3
Allocation Unit Size: 279.2GB
Allocation Unit Data Size for RAID4: 186.2GB
Allocation Unit Data Size for RAID-DP: 93.08GB
Allocation Unit Data Size for RAID-TEC: -
Storage Type: SSD
Storage Pool Usable Size: 0B
Storage Pool Total Size: 1.09TB
Is Pool Healthy?: true
State of the Storage Pool: normal
Reason for Storage Pool Being Unhealthy: -
Job ID of the Currently Running Operation: -
cluster2::> storage pool show-aggregate -storage-pool ClusterCache -instance
Name of Storage Pool: ClusterCache
Aggregate: sata_data_1
Capacity: 558.5GB
Number of AU's Assigned to This Aggregate: 2
Original Owner Name: cluster2-node01
Node: cluster2-node01
Name of Storage Pool: ClusterCache
Aggregate: sata_data_2
Capacity: 558.5GB
Number of AU's Assigned to This Aggregate: 2
Original Owner Name: cluster2-node02
Node: cluster2-node02
2 entries were displayed.
So, I have two nodes, two aggregate set as hybrid, 4 SSD 960GB:
storage aggregate modify -aggregate cluster4_01_NL_SAS_1 -hybrid-enabled true
storage aggregate modify -aggregate cluster4_02_NL_SAS_1 -hybrid-enabled true
I will create a storage pool with 3 SSD, so one remains as spare, and assign 2 AU per aggregate with raid4 type:
storage aggregate add cluster4_01_NL_SAS_1 -storage-pool StoragePool_c4_Cache1 -allocation-units 2 -raidtype raid4
storage aggregate add cluster4_02_NL_SAS_1 -storage-pool StoragePool_c4_Cache1 -allocation-units 2 -raidtype raid4
I prefer raid4 instead of Raid-DP in order to have more space on flash-pool.
All should works fine, correct? Any suggestions?
Thanks
You are looking at two methods for FlashPool. The old method used WHOLE SSDs. The NEW method uses PARTITIONED SSDs.
All 4 SSDs should be modified to be "owned" by the same node.
Create the storage pool with 3 x SSDs.
Modify the aggr to be hybrid enabled
Finally "add-disks" to each of your two AGGRs. Like you mention, use a storage-pool and specify raid4 to maximize usage.
You can only do one at a time. You need to wait for the first to finish before doing the second aggr. It will complain and not run if it is still adding-cache.
Ok, after set as 2 the number of AU per aggregate I have add successfully the flash-pool to the two aggregates.
Now ONTAP but also AIQ warning me about a low number of SSD spare on node01, since of 4 SSD there is only one spare assigned to node02:
cluster4::> storage pool show-disks
Storage Pool Name: StoragePool_c4_Cache1
Disk
Disk Type Usable Size Total Size
-------- ------- ----------- -----------
1.0.0 SSD 894.0GB 894.3GB
1.0.1 SSD 894.0GB 894.3GB
1.0.10 SSD 894.0GB 894.3GB
3 entries were displayed.
cluster4::> storage aggregate show-spare-disks -disk-type SSD
Original Owner: cluster4-02
Pool0
Spare Pool
Usable Physical
Disk Type Class RPM Checksum Size Size Status
---------------- ------ ----------- ------ -------------- -------- -------- --------
1.0.11 SSD solid-state - block 894.0GB 894.3GB zeroed
Should I change the assignment of this SSD or in case of one SSD fail, indipendentment from node01 or node02, this spare will use?
I remember that in the past only one SSD for spare, in Flash Pool, is safe for all nodes in HA.
Thanks