Failover event cause-and-effect table

Failover events cause a controller failover in HA pairs. The storage system responds differently depending on the event and the type of HA pair.

Cause-and-effect table for standard or mirrored HA pairs

Event Does the event trigger failover? Does the event prevent a future failover from occurring, or a failover from occurring successfully? Is data still available on the affected volume after the event?
Single storage system Standard or mirrored HA pair
Single disk failure No No Yes Yes
Double disk failure (2 disks fail in same RAID group) Yes, unless you are using SyncMirror or RAID-DP, then no Maybe. If root volume has double disk failure, or if the mailbox disks are affected, no failover is possible No, unless you are using RAID-DP or SyncMirror, then yes No, unless you are using RAID-DP or SyncMirror, then yes
Triple disk failure (3 disks fail in same RAID group) Maybe. If SyncMirror is being used, no takeover occurs; otherwise, yes Maybe. If root volume has triple disk failure, no failover is possible No No
Single HBA (initiator) failure, Loop A Maybe. If SyncMirror or multipath HA is in use, then no; otherwise, yes Maybe. If root volume has double disk failure, no failover is possible Yes, if multipath HA or SyncMirror is being used Yes, if multipath HA or SyncMirror is being used, or if failover succeeds
Single HBA (initiator) failure, Loop B No Yes, unless you are using SyncMirror or multipath HA and the mailbox disks are not affected, then no Yes, if multipath HA or SyncMirror is being used Yes, if multipath HA or SyncMirror is being used, or if failover succeeds
Single HBA initiator failure (both loops at the same time) Yes, unless the data is mirrored on a different (up) loop or multipath HA is in use, then no takeover needed Maybe. If the data is mirrored or multipath HA is being used and the mailbox disks are not affected, then no; otherwise, yes No, unless the data is mirrored or multipath HA is in use, then yes No failover needed if data is mirrored or multipath HA is in use
AT-FCX failure (Loop A) Only if multidisk volume failure or open loop condition occurs, and neither SyncMirror nor multipath HA is in use Maybe. If root volume has double disk failure, no failover is possible No Yes, if failover succeeds
AT-FCX failure (Loop B) No Maybe. If SyncMirror or multipath HA is in use, then no; otherwise, yes Yes, if multipath HA or SyncMirror is in use Yes
IOM failure (Loop A) Only if multidisk volume failure or open loop condition occurs, and neither SyncMirror nor multipath HA is in use Maybe. If root volume has double disk failure, no failover is possible No Yes, if failover succeeds
IOM failure (Loop B) No Maybe. If SyncMirror or multipath HA is in use, then no; otherwise, yes Yes, if multipath HA or SyncMirror is in use Yes
Shelf (backplane) failure Only if multidisk volume failure or open loop condition occurs, and data isn’t mirrored Maybe. If root volume has double disk failure or if the mailboxes are affected, no failover is possible Maybe. If data is mirrored, then yes; otherwise, no Maybe. If data is mirrored, then yes; otherwise, no
Shelf, single power failure No No Yes Yes
Shelf, dual power failure Only if multidisk volume failure or open loop condition occurs and data is not mirrored Maybe. If root volume has double disk failure, or if the mailbox disks are affected, no failover is possible Maybe. If data is mirrored, then yes; otherwise, no Maybe. If data is mirrored, then yes; otherwise, no
Controller, single power failure No No Yes Yes
Controller, dual power failure Yes Yes, until power is restored No Yes, if failover succeeds
HA interconnect failure (1 port) No No Not applicable Yes
HA interconnect failure (both ports) No Yes Not applicable Yes
Ethernet interface failure (primary, no interface group) Yes, if set up to do so No Yes Yes
Ethernet interface failure (primary, interface group) Yes, if set up to do so No Yes Yes
Ethernet interface failure (secondary, interface group) Yes, if set up to do so No Yes Yes
Ethernet interface failure (interface group, all ports) Yes, if set up to do so No Yes Yes
Tape interface failure No No Yes Yes
Heat exceeds permissible amount Yes No No No
Fan failures (disk shelves or controller) No No Yes Yes
Reboot Yes No No Yes, if failover occurs
Panic Yes No No Yes, if failover occurs

Cause-and-effect table for stretch and fabric-attached MetroClusters

Event Does the event trigger failover? Does the event prevent a future failover from occurring, or a failover from occurring successfully? Is data still available on the affected volume after the event?
Stretch MetroCluster Fabric Attached MetroCluster
Single disk failure No No Yes Yes
Double disk failure (2 disks fail in same RAID group) No No Yes Yes, with no failover necessary
Triple disk failure (3 disks fail in same RAID group) No No Yes, with no failover necessary Yes, with no failover necessary
Single HBA (initiator) failure, Loop A No No Yes Yes, with no failover necessary
Single HBA (initiator) failure, Loop B No No Yes Yes, with no failover necessary
Single HBA initiator failure, (both loops at the same time) No No Yes, with no failover necessary Yes, with no failover necessary
AT-FCX failure (Loop A) No No Yes Not applicable
AT-FCX failure (Loop B) No No Yes Not applicable
ESH4 or IOM Failure (loop A) No No Yes Yes
ESH4 or IOM Failure (loop B) No No Yes Yes
Shelf (backplane) failure No No Yes Yes, with no failover necessary
Shelf, single power failure No No Yes Yes
Shelf, dual power failure No No Yes Yes, with no failover necessary
Controller, single power failure No No Yes Yes
Controller, dual power failure Yes Yes, until power is restored Yes, if failover succeeds Yes, if failover succeeds
HA interconnect failure (1 port) No No Yes Yes
HA interconnect failure (both ports) No No; failover is possible Yes Yes
Ethernet interface failure (primary, no interface group) Yes, if set up to do so No Yes Yes
Ethernet interface failure (primary, interface group) Yes, if set up to do so No Yes Yes
Ethernet interface failure (secondary, interface group) Yes, if set up to do so No Yes Yes
Ethernet interface failure (interface group, all ports) Yes, if set up to do so No Yes Yes
Tape interface failure No No Yes Yes
Heat exceeds permissible amount Yes No Yes, if failover occurs Yes, if failover occurs
Fan failures (disk shelves or controller) No No Yes Yes
Reboot Yes No Maybe. Depends on cause of reboot Maybe. Depends on cause of reboot
Panic Yes No Maybe. Depends on cause of panic Maybe. Depends on cause of panic