AFF
AFF
Hello,
could someone please help clarify if that documentation applies to root-data-data partitioning?
Do I have to leave 2 different ssd with 2 spare partitions (root+data) for the HA pair?
Here's what i have on one of our AFF
cluster::> storage aggregate show-spare-disks
Original Owner: cluster-01
Pool0
Root-Data1-Data2 Partitioned Spares
Local Local
Data Root Physical
Disk Type Class RPM Checksum Usable Usable Size Status
---------------- ------ ----------- ------ -------------- -------- -------- -------- --------
1.0.7 SSD solid-state - block 0B 37.42GB 3.49TB zeroed
1.0.19 SSD solid-state - block 1.73TB 0B 3.49TB zeroed
Original Owner: cluster-02
Pool0
Root-Data1-Data2 Partitioned Spares
Local Local
Data Root Physical
Disk Type Class RPM Checksum Usable Usable Size Status
---------------- ------ ----------- ------ -------------- -------- -------- -------- --------
1.0.19 SSD solid-state - block 1.73TB 37.42GB 3.49TB zeroed
3 entries were displayed.
Thank you
Lorenzo
Solved! See The Solution
If all your drives are RD2 (Root-Data-Data) partitioned, then they all show as "shared".
Technically speaking one drive on either controller can be used as a spare on both. Why?
If Any drive fails, it fails on both nodes. This causes a reconstruction on both nodes since the drive is partitioned between both. If there is a WHOLE (unpartitioned) drive, it is automatically partitioned to match what is there and reconstruction occurs on all three partitions (Root/Data/Data).
If you have a mix of RD2 and whole SSDs (unpartitioned), then you should leave at least one unpartitioned drive per controller.
If all your drives are RD2 (Root-Data-Data) partitioned, then they all show as "shared".
Technically speaking one drive on either controller can be used as a spare on both. Why?
If Any drive fails, it fails on both nodes. This causes a reconstruction on both nodes since the drive is partitioned between both. If there is a WHOLE (unpartitioned) drive, it is automatically partitioned to match what is there and reconstruction occurs on all three partitions (Root/Data/Data).
If you have a mix of RD2 and whole SSDs (unpartitioned), then you should leave at least one unpartitioned drive per controller.
Sorry,
maybe I have not explained very well...
In my example, do is better to align spare partitions in order to have all spares on 1.0.19? (of course better to have 2 full expensive spare drives) 🙂
It's now clear that the mentioned doc applies to root-data and not to root-data-data partitioning.
Thanks a lot