How does ONTAP quorum loss affect SAN services?
Applies to
ONTAP 9
Answer
- For an explanation of how quorum and epsilon work in ONTAP, see About quorum and epsilon.
- Without cluster quorum, cluster management is impossible and aggregates and volumes will report unknown status.
- This includes inability to manage LUN map/unmap, creation or modification of igroups, and other such administrative tasks.
- However, as long as scsiblade does not go out of quorum and at least one node survives in HA pair in takeover, SAN data will continue being served, even if the cluster itself has lost quorum and cluster applications are offline.
- Initiator logins, both existing ones and new ones, will work, as long as they're part of existing igroups.
- If scsiblade does go out of quorum (resulting in SAN outage), this is reported in EMS:
[cluster-01: ThreadHandlerun: scsiblade.out.of.quorum:EMERGENCY]: This node is no longer in SAN quorum and will not provide any SAN service until cluster communications are restored.
- Loss of scsiblade quorum will also trigger an AutoSupport, see SCSIBLADE SERVICE UNAVAILABLE
- Scsiblade back in quorum/initialization of scsiblade services on boot up, will report the following in EMS:
[cluster-01: BcomkaThr_02: scsiblade.online:notice]: This node has completed initialization of the SAN configuration cache. FCP and iSCSI traffic are enabled on this node.
Additional Information
additionalInformation_text