This release note discusses the new features, known limitations, and open caveats for Cisco UCS Virtual Interface Card (VIC) Drivers, Release 2.2. It does not cover new features, resolved and open caveats for third-party drivers, and known limitations.
Note We sometimes update the documentation after original publication. Therefore, you should also review the documentation on Cisco.com for any updates.
Use this document in conjunction with the document listed in the Related Documentation section.
Table 1 shows the online change history for this document.
Table 2 lists the caveats that have been resolved since Release 2.2.
Table 2 Resolved Caveats
The data corruption issue seen with controller resets, controller panic, and target port blocks is resolved on ESXi hosts running 3GFI and 2GFI setups with NetApp.
The issue with the ESXi 6.0 host running 3GFI is no longer terminating during fNIC traffic tests.
The issue with the connection to a storage target getting lost when a FC link is removed from the SAN port-channel is resolved.
ESXi 5.1 U3
The issue with the ESXi 5.1 U3 host with fNIC 184.108.40.206. is no longer terminating.
The fNIC drivers are no longer crashing when the multi-queue block IO queueing mechanism (blk-mq) is enabled in the Unbreakable Enterprise Kernel Release 4 (UEK4).
The XEN server panic issue when installing the XEN6.5 RC2 through SAN boot is resolved.
The issue with the UCS Manager running ESXi 5.1 and fNIC drivers is no longer terminating with stacktraces.
The fNIC driver name and meta data must be different across RHEL versions.
fNIC version 220.127.116.11
The issue when running the vfcshut and noshut commands for several hours on Lex adapters is no longer causing an ESXi host panic with fNIC_fc_trace_set_data () at the top of the stack.
The IOMMU fault no longer occurs when IO and abort IO is out of order.
The fNIC drivers are no longer crashing with a null pointer value.
When the exchange ID recycle is turned off, exchange IDs are no longer being reassigned prematurely for new commands under extreme I/O load.
The probing issue with at least one missing vNIC after each reboot of the server is resolved.
The eNIC drivers are no longer issuing a Warning at lib/dma-debug.c:937 check_unmap+0x47b/0x920().
The hardware csum failure on Cisco hardware is resolved.
UCS Mini running on ESXi is now able to accept packets that are less than 60 bytes.
The Cisco UCS VIC 1240 adapter driver 18.104.22.168 can now allow the Fabric Interconnect to discover the MAC address.
The issue with packets that should be untagged coming across RHEL 7as being tagged, with VLAN 0 is resolved.
The ESXi operating system kernel is no longer crashing during certain operations, such as updating the VIC adapter firmware while running VM-FEX.
A Cisco UCS B420 M3 blade server configured with more than 1.1 TB of memory and using a Cisco VIC adapter fails to render the multipath file system (FS) in read-only mode after performing a read or write operation.
Windows Driver 22.214.171.124
Windows iSCSI crash dump driver installation no longer fails on a non-English locale.
A VIC management driver no longer causes a Windows OS 2008 R2 install failure.
The fNIC driver no longer fails to update in Linux.
* where x is the action to take. Set it to 1 to create the event log, but the system continues working without BSOD.
Note Following are the four HangRecoveryAction settings for CSCuj20256:
0 - Disables the heartbeat and monitoring mechanism.
1 - Logs an event in the system log of the Event Viewer.
2 - Terminates the Cluster Service.
3 - Causes a Stop error (Bugcheck) on the cluster node. <<-- default.
Customers using Red Hat Enterprise Linux version 5.8 or an earlier version should be aware that a libfc bug exists in distributions. The details of this bug are available in the Red Hat Bugzilla 884740.
Known Limitations and Behaviors
The following limitations were introduced with support of ARFS in Cisco UCS Manager release 2.2(2c):
Supported only on UCS-VIC-M82-8P and UCSB-MLOM-40G-01.
Supports 64 flow classifier entries per vNIC.
Supports approximately 992 flow classifier entries per adaptor (approximately 15 vNICs with ARFS enabled).
The following limitations were introduced with support of AIC in Linux ENIC driver in Cisco UCS Manager release 2.2(2c):
Does not provide any advantage when the link utilization is greater than 80%.
When enabled, this feature disables static coalescing.
The following limitations were introduced with support of NetQueue in Cisco UCS Manager release 2.2(2c):
VMware recommends no more than 8 NetQueues per port for standard or jumbo (9000) frame configurations.
– When enabling NetQueue on vNICs, Cisco recommends using the default VMware adapter policy.
– Enabling too many NetQueues on too many vNICs can result in the host exceeding limits for netPKtheap and msix interrupts.
NetQueue should be enabled only for vNICs configured to use MSI-X interrupts.
Cisco and the Cisco logo are trademarks or registered trademarks of Cisco and/or its affiliates in the U.S. and other countries. To view a list of Cisco trademarks, go to this URL: www.cisco.com/go/trademarks. Third-party trademarks mentioned are the property of their respective owners. The use of the word partner does not imply a partnership relationship between Cisco and any other company. (1110R)
Any Internet Protocol (IP) addresses and phone numbers used in this document are not intended to be actual addresses and phone numbers. Any examples, command display output, network topology diagrams, and other figures included in the document are shown for illustrative purposes only. Any use of actual IP addresses or phone numbers in illustrative content is unintentional and coincidental.