DS2246 disk shelf failures
- Views:
- 3,336
- Visibility:
- Public
- Votes:
- 2
- Category:
- disk-shelves
- Specialty:
- hw
- Last Updated:
- 9/30/2021, 7:19:25 PM
Applies to
DS2246 (EOA), FAS2552 (EOA), FAS2240-2 (EOS)
Description
-
A DS2246 SAS2 storage disk shelf might experience a failure where the shelf loses all power.
-
If a DS2246 is used as an external storage shelf, refer to the following table to determine if there will be a multi-disk panic (determined by the number of drives per RAID group that will fail if a DS2246 which contains up to 24 drives were to lose power). These drives might be distributed in various ways, from 1-2 per RAID group to all 24 within a single RAID group.
| Criteria | RAID-DP/RAID-TEC Alone | RAID-DP/RAID-TEC with SyncMirror |
|---|---|---|
| Failures protected against |
|
|
| Failures not protected against |
|
Enough concurrent failures to defeat protection by RAID-DP or RAID-TEC, simultaneously occurring on both mirrored plexes |
-
In the following example, Shelf 2 suddenly loses all power and does not show up in
Sysconfig -a:
Shelf 0: DS2246 Firmware rev. IOM6 A: 0173 IOM6 B: 0173
Shelf 1: DS2246 Firmware rev. IOM6 A: 0173 IOM6 B: 0173
Shelf 3: DS2246 Firmware rev. IOM6 A: 0173 IOM6 B: 0173
Note: Shelf 2 is missing.
-
The following alert might be seen from ONTAP if the system panics in accordance with the table, above:
Panic string: aggr con_stor_01_aggr1: raid volfsm, fatal multi-disk error. raid type raid_dp in SK process config_thread
-
It is followed by the following “
Dual Path to Disk Shelf” message:
(Health Monitor process nchm: DualPathToDiskShelf_Alert[50:0a:09:80:00:81:33:8c]) ERROR
-
Power supplies still show amber lights; however, there is no disk activity or LED lights in front of the impacted disk shelf.
-
If a DS2246 is used as part of a two-in-a-chassis storage controller configuration (e.g. FAS2552, FAS2240-2), a full shelf power failure will manifest as an unresponsive and unavailable HA pair. Storage owned by that HA pair will be inaccessible. The only available protection against this is ensuring that a disaster recovery solution is in place, e.g., MetroCluster or SnapMirror Synchronous.
Note: The impact statement for a two-in-a-chassis configuration applies to any two-in-a-chassis configuration and not just those that reside in a DS2246-based chassis.
