During a scheduled NetApp ONTAP upgrade (or LIF migration), multiple critical production NFS and CIFS storage shares abruptly went offline immediately after a storage failover. Clients completely lost network connectivity to the storage Logical Interfaces (LIFs), even though the NetApp cluster reported that all networks and ports were completely healthy and active.
Pktt/tcpdump:
The trace confirmed that the NetApp cluster was correctly broadcasting Gratuitous ARP (GARP) packets to update the network fabric. However, Arista interfaces were dropping the packets instead of processing them.
Context:
When a NetApp Logical Interface (LIF) migration or storage failover/giveback occurs, the NetApp cluster broadcasts Gratuitous ARP (GARP) packets for approximately 10 seconds to update upstream network switches with the LIF's new MAC address location.
If you are using Arista Switches, the LIF will frequently stop responding or experience a complete ping timeout because by default, Arista EOS interfaces reject/drop gratuitous ARP request packets. Because the switch drops these packets, its ARP and MAC tables are not updated, and traffic continues to be routed to the old, inactive physical port or node.
Key Symptoms:
LIF Unreachable: Clients instantly lose access to CIFS/NFS shares or iSCSI datastores following a LIF migration, storage failover, or automated ONTAP upgrade.
Ping Failures: Pings fail from the NetApp LIF to its local gateway, and downstream switches show no newly learned MAC info.
Cause:
Disabled by Default: Arista EOS explicitly drops incoming gratuitous ARP requests unless an explicit overriding configuration is applied to the Layer 3 interface, Layer 3 port channel, or SVI (Switch Virtual Interface).
Temporary Workaround Effect:
Temporarily migrating the LIF back to its original port or manually flushing/clearing the switch's local ARP cache instantly restores connectivity
NetApp explicitly documents this exact issue in Official NetApp KB Articles, confirming that external network switches dropping GARP packets during failover, giveback, or manual migration causes LIF timeouts. You can reference the following official documentation for details and troubleshooting steps:
https://kb.netapp.com/on-prem/ontap/da/NAS/NAS-KBs/LIF_unreachable_after_giveback_due_to_ARP_table_not_updated_on_the_switch
https://kb.netapp.com/on-prem/ontap/da/NAS/NAS-KBs/LIF_not_responding_after_being_migrated_or_failing_over_to_another_node
Comments
Post a Comment