- ONTAP 9
- Clustered Data ONTAP 8.3
- In ONTAP 9, while adding a shelf from a different cluster (or a 7-Mode system) to a node, aggregates on that shelf may be flagged as foreign and the drives belonging to that aggregate as orphan disks.
- This issue can also be encountered after replacing a failed disk which was never zeroed.
- Aggregate will report failed and foreign under the
aggr statuscommand and EMS may report RAID assimilation failures during reboot/takeover
aggr0(1) failed raid_dp, aggr diskroot, raidtype=raid_dp, raidsize=24,
foreign raid_lost_write=on, resyncsnaptime=5,
[?] Fri Jul 16 06:01:45 CDT [cluster1-01: config_thread: raid.assim.rg.missingChild:debug]: Aggregate partner_foreign:aggr0, rgobj_verify: RAID object 0 has only 5 valid children, expected 10.
[?] Fri Jul 16 06:01:45 CDT [cluster1-01: config_thread: raid.assim.plex.missingChild:debug]: Aggregate partner_foreign:aggr0, plexobj_verify: Plex 0 only has 0 working RAID groups (1 total) and is being taken offline
[?] Fri Jul 16 06:01:45 CDT [cluster1-01: config_thread: raid.assim.mirror.noChild:debug]: Aggregate partner_foreign:aggr0, mirrorobj_verify: No operable plexes found.
[?] Fri Jul 16 06:01:45 CDT [cluster1-01: cf_takeover: raid.fm.takeoverFail:alert]: RAID takeover failed: Partner label processing failed.
- This is caused due to the disks not being zeroed (and containing the label) and therefore retaining the aggregate information from the earlier system.
- These 'foreign' aggregates are visible only when using the node shell.
- Accessing the foreign volumes inside these aggregates on the clustered Data ONTAP system to which the shelves have been migrated is not supported.
- However, it is supported to remove the foreign aggregates, turning the relevant drives into unzeroed spares.