Write latency spike seen at VM after invoking AutoSupport
Applies to
- ONTAP 9
- VMware ESXi 8.0.1
- Cisco UCS
Issue
- Around 3 minutes after AutoSupport is invoked (
system node autosupport invoke -type all -node *) there is a write latency spike up to 1 second seen at the VM in vSphere MANIFEST.XMLof the AutoSupport reports a timeout forfcp-topology-switches.xml:
Timeout: Operation "fct_topology_ie_iterator::next_imp()" took longer than 60 seconds to complete [from mgwd on node "STGT01-C05B" (VSID: -1) to kernel at 127.0.0.1]
- There are processing failures of
SLI_IO_FC_CTcommands in EMS after AutoSupport is invoked:
FC port 2d has a WQE processing failure of type SLI_IO_FC_CT, command GIELN, subtype x1, with status of 3 and an extended status x2 for NPIV port 0.FC port 2b has a WQE processing failure of type SLI_IO_FC_CT, command GPL, subtype x1, with status of 3 and an extended status x2 for NPIV port 0.- Issue remains the same after hardware replacements and trying different ports on switch and HBA
- ESXi host
vmkernel.logshows issues with NFNIC after AutoSupport is invoked:
nfnic: <4>: INFO: fnic_taskMgmt: 2128: TaskMgmt abort sc->cdb: 0x2a
nfnic: <4>: INFO: fnic_abort_cmd: 3793: Abort cmd called for Tag: 0x26b issued time: 180752 ms CMD_STATE: FNIC_IOREQ_CMD_PENDING CDB Opcode: 0x2a sc:0x45b91f0b9140 flags: 0x3 lun: 1 target: 0x161c02
WARNING: nfnic: <4>: fnic_abort_cmd: 3808: Abort for cmd tag: 0x26b in pending state
