e0a/e0b link flaps on A300/FAS8200, A200/FAS2600, A220/FAS2700,C190 may cause a Takeover
- Views:
- 24,704
- Visibility:
- Public
- Votes:
- 15
- Category:
- fas-systems
- Specialty:
- hw
- Last Updated:
- 11/25/2024, 7:18:29 PM
Applies to
- AFF A300, FAS8200
- AFF A200, FAS2650, FAS2620
- AFF A220, AFF C190, FAS2750, FAS2720
- ONTAP 9
Issue
- Cluster ports e0a or e0b (or both ports) experience link flaps or go down at the same time.
Tue Oct 03 11:08:31 CEST [node1: ixgbe/e0b: snmp.link.down:info]: Interface 2 is down.
Tue Oct 03 11:08:31 CEST [node1: ixgbe/e0b: netif.linkDown:info]: Ethernet e0b: Link down, check cable.
Tue Oct 03 11:08:31 CEST [node1: ixgbe/e0a: snmp.link.down:info]: Interface 1 is down.
Tue Oct 03 11:08:31 CEST [node1: ixgbe/e0a: netif.linkDown:info]: Ethernet e0a: Link down, check cable.
Tue Oct 03 11:08:32 CEST [node2: ixgbe/e0b: snmp.link.down:info]: Interface 2 is down.
Tue Oct 03 11:08:32 CEST [node2: ixgbe/e0b: netif.linkDown:info]: Ethernet e0b: Link down, check cable.
Tue Oct 03 11:08:32 CEST [node2: ixgbe/e0a: snmp.link.down:info]: Interface 1 is down.
Tue Oct 03 11:08:32 CEST [node2: ixgbe/e0a: netif.linkDown:info]: Ethernet e0a: Link down, check cable.
- Check cluster port status and storage failover status:
cluster1::> network port show -ipspace Cluster
Node: cluster1-01
Speed(Mbps) Health
Port IPspace Broadcast Domain Link MTU Admin/Oper Status
--------- ------------ ---------------- ---- ---- ----------- --------
e0a Cluster Cluster down 9000 1000/- -
e0b Cluster Cluster down 9000 1000/- -
Node: cluster1-02
Speed(Mbps) Health
Port IPspace Broadcast Domain Link MTU Admin/Oper Status
--------- ------------ ---------------- ---- ---- ----------- --------
e0a Cluster Cluster down 9000 1000/- -
e0b Cluster Cluster down 9000 1000/- -
4 entries were displayed.
cluster1::> storage failover show
Takeover
Node Partner Possible State Description
------------- -------------- -------- -------------------------------------
cluster1-01 cluster1-02 false Connected to cluster-02, Partial
giveback, Takeover is not possible:
The version of software running on
each node of the SFO pair is
incompatible, NVRAM log not synchronized
cluster1-02 cluster1-01 - Waiting for cluster applications to
come online on the local node
Offline applications: mgmt, vldb,
vifmgr, bcomd, crs.
If the ports do not come back up and if Connectivity, Liveliness and Availability Monitor (CLAM) is enabled
- An "out of quorum" panic will occur on one of the nodes.
PANIC : Received PANIC packet from partner, receiving message is (Coredump and takeover initiated because Connectivity, Liveliness and Availability Monitor (CLAM) has determined this node is out of quorum.
- The node that panics will be taken over and the surviving node will be serving all data.
If the ports do not come back up and if Connectivity, Liveliness and Availability Monitor (CLAM) is NOT enabled
- There will not be a storage takeover and both nodes will go out of quorum. Neither node will be serving data.
- See: SU436: [Impact: Critical] CLAM takeover default configuration changed
- Similar messages can be found in EMS log:
Jun 08 12:30:09 [xxx-02:vifmgr.clus.linkdown:EMERGENCY]: The cluster port e0b on node naptp06c-02 has gone down unexpectedly.
Jun 08 12:30:10 [xxxc-02:vifmgr.clus.linkdown:EMERGENCY]: The cluster port e0a on node naptp06c-02 has gone down unexpectedly.
Jun 08 12:31:00 [xxx-02:monitor.globalStatus.critical:EMERGENCY]: Controller failover of xxx-01 is not possible: partner mailbox disks not accessible or invalid. One or more mirrored aggregates are degraded.
Jun 08 12:31:02 [xxx:callhome.clam.node.ooq:EMERGENCY]: Call home for NODE(S) OUT OF CLUSTER QUORUM.
