The following NFNIC messages are seen in the vmkernel.log:
WARNING: nfnic: <4>: fnic_fcpio_icmnd_cmpl_handler: 1636: sc: 0x459b12896d00 tag: 0x4e1 hdr status: FCPIO_ABORTED IO failure!
WARNING: nfnic: <4>: fnic_fcpio_icmnd_cmpl_handler: 1636: sc: 0x459b128ea7c0 tag: 0x507 hdr status: FCPIO_ABORTED IO failure!
WARNING: nfnic: <4>: fnic_fcpio_icmnd_cmpl_handler: 1636: sc: 0x459b1280c880 tag: 0x533 hdr status: FCPIO_ABORTED IO failure!
This may also lead to VMFS Metadata On-Disk corruption and/or corruption within the GuestOS of VMs
CTIO errors or Out of Order frames may be apparent in the SAN. Read/Writes may fail, but be marked as successful and can cause data inconsistency on remote storage datastores.
Cisco has released an updated Native FNIC driver (4.0.0.65) that addresses the FCP abort issue and prevents the lost write situation that leads to corruption.
For more information reference the CISCO defect CSCvx64370
If the environment meets the above criteria, until the long term fix is available from Cisco, we recommend upgrading as soon as possible to maintain data integrity. The issue can be mitigated by upgrading to Purity version 5.3.15+, 6.0.6+, and 6.1.3+.
Pure Storage Support will be happy to assist customers with this upgrade.