Table Of Contents
Release Notes for Cisco UCS Release 1.0(1e)
Date: August 1, 2009Part Number: OL-20090-01 B0
This document describes the new features, system requirements, and caveats for Cisco UCS, Release 1.0(1e). Use this document in conjunction with the documents listed in the "Related Documentation" section.
Note We sometimes update the documentation after original publication. Therefore, you should also review the documentation on Cisco.com for any updates.
Note Table 1 shows the online change history for this document.
This document includes the following sections:
The Cisco® Unified Computing System is a next-generation data center platform that unites compute, network, storage access, and virtualization into a cohesive system designed to reduce total cost of ownership (TCO) and increase business agility. The system integrates a low-latency, lossless 10 Gigabit Ethernet unified network fabric with enterprise-class, x86-architecture servers. The system is an integrated, scalable, multichassis platform in which all resources participate in a unified management domain.
The Cisco UCS client must meet or exceed the following minimum system requirements:
•The Cisco UCS Manager GUI is a Java-based application that requires Sun JRE 1.6 or later.
•The UCS Manager uses webstart and supports the following web browsers:
–Microsoft Internet Explorer 6.0 or higher
–Mozilla Firefox 3.0 or higher
•The UCS Manager is supported on the following operating systems:
–Microsoft Windows XP
–Microsoft Windows Vista
–-Red Hat Enterprise Linux 5.0 or higher
Updating the Cisco UCS Firmware
To update the Cisco UCS firmware, refer to the "Firmware Management" chapter in either of the following documents:
•Cisco UCS Manager CLI Configuration Guide located at http://www.cisco.com/en/US/docs/unified_computing/ucs/sw/cli/config/guide/CLI_Config_Guide..pdf
•Cisco UCS Manager GUI Configuration Guide located at http://www.cisco.com/en/US/docs/unified_computing/ucs/sw/gui/config/guide/GUI_Config_Guide..pdf
Hardware and Software Interoperability
For detailed information about storage switch, operating system, adapter, adapter utility, and storage array interoperability, see the Hardware and Software Interoperability Matrix located at:
Known Limitations and Behaviors
UCS Manager release 1.0(1e) supports up to five 5108 Chassis', or up to 40 B00 servers.
(CSCtd32884 and CSC71310) The type of adapter in a server impacts the maximum MTU supported. Network MTU above the maximums may cause the packet to be dropped for the following adapters:
· The Cisco UCS CNA M71KR adapter supports a maximum MTU of 9216.
· The Cisco UCS 82598KR-CI adapter supports a maximum MTU of 14000.
Recovering the Corrupt BIOS on a Server
(CSCta55689) Before you attempt to recover a corrupt BIOS, remove all attached or mapped USB storage from a server before you attempt to recover the corrupt BIOS on that server. If an external USB drive is attached or mapped from vMedia to the server, BIOS recovery fails.
(CSCtc70962) Choose a strong password with at least one capital letter and one number. This password cannot be blank. If a blank password is created, the account will become unusable.
Changing Management IP Port
(CSCtb42606) After the initial configuration, you can change the management IP port and the related subnet mask in the Cisco UCS Manager CLI. You cannot make this change in the Cisco UCS Manager GUI.
The following known issues will be corrected in future versions of these documents.
Incorrect VLAN IDs Range
The USCMGUI and CLI configuration guide sections on named VLANs specified an incorrect reserved VLAN ID range. The text read: "You cannot create VLANs with IDs from 3968 to 4048. This range of VLAN IDs is reserved." The correct range is 3968 to 4047.
Error in upgrade sequence
The USCM GUI and CLI configuration guide sections regarding upgrade sequences has "Updating and Activating the Firmware on a Fabric Interconnect" before "Updating and Activating the Cisco UCS Manager Software" which is incorrect. The correct sequence should be "Updating and Activating the Cisco UCS Manager Software" and then "Updating and Activating the Firmware on a Fabric Interconnect."
The documentation was updated to:
•Clarify that a user can log in with any Cisco UCS account to view the firmware versions.
•Emphasize the recommendation that additional user accounts be created on a Cisco UCS instance (already in the documentation) and that users not rely on having only the superuser admin account created when the fabric interconnects are configured.
•Add a recommendation that, if no additional accounts are created, that the users note down and save the firmware versions outside of the Cisco UCS instance.
•Add examples of the firmware versions and updated the installables path (replacing 'fabric' with 'switch' in the procedures.
The documentation was updated to add information about Admin password requirements to the following topics:
•Performing the Initial System Setup for a Standalone Fabric Interconnect
•Performing the Initial System Setup for the First Fabric Interconnect
•Recovering an Admin Password for a Standalone Configuration
•Recovering an Admin Password for a Cluster Configuration
This section lists the open caveats for this release.
Symptom Local user passwords can not contain "$" character.
Workaround Do not include the "$" character in local user passwords.
Symptom If an admin failover is performed in a loop multiple times, Telnet and SSH connections to the virtual IP address are not immediately cleared. This can cause the UCS Manager to reach the maximum session limit, which prevents new sessions from being initiated.
Workaround Avoid performing admin failovers in a loop.
Symptom The output of keyring commands in the show configuration command is not in the expected sequence. The keyring commands should be performed in specific order for the keyring to be successfully created.
Workaround Manually execute keyring commands in expected sequence.
Symptom The default keyring allows you to set a trusted point name.
Workaround This is a cosmetic issue. Do not set trusted point name for default keyring.
Symptom When using a fully qualified domain name (FQDN) of an LDAP provider, the FQDN is not resolved with DNS, and user authentication using the LDAP provider does not work.
Workaround Use an IP address instead of an FQDN when creating LDAP providers.
Symptom The IPMI user database in not backed up in the UCS Manager when you export a configuration backup.
Workaround After the configuration import is done, manually configure the IPMI user profile.
Symptom When configuring an LDAP provider, if the distinguished name (DN) contains a space, NX-OS fails to apply the configuration.
Workaround Do not use DNs that contain spaces when configuring an LDAP provider.
Symptom You cannot qualify a server adapter by selecting non-virtualized-eth that does not have support protection. This applies to the UCS 82598KR-CI 10-Gigabit Ethernet Adapter.
Workaround Create a qualifier based on eth-capability and select protection as an add on.
Symptom The UCS Manager CLI and GUI have the following terminology differences for adapter policies:
1. They are called profiles in the CLI and policies in the GUI
2. The fc-profile work queue is called work-queue in the CLI and Transmit queue in the GUI.
Workaround Note the terminology differences.
Symptom With fabric failover being enabled in a service profile associated with a QLogic or Emulex CNA, there is a potential traffic loss in the network-to-host direction if the fabric interconnect is operating in End Host mode. The duration of loss depends on how soon after failover the host sends a packet towards the network.
Workaround The traffic loss ends as soon as the host sends a packet after the failover. If the host is a silent host, inject traffic from the host towards the network.
Symptom When a service profile containing two vNICs and having failover enabled is applied to QLogic or Emulex CNAs, the failback timeout specified in the adapter policy for the second vNIC has no effect. The failback timeout specified in the adapter policy and applied to the first vNIC is applied to the whole adapter and is effective for both vNICs.
Workaround Specify the desired failback timeout in the adapter policy and apply to the first vNIC.
Symptom Broadcast packets do not reach the server, even when promiscuous mode is enabled on the server. The IGMP Join group is seen on fabric connect, but multicast packets do not reach the server for multicast groups added after the multicast table on the adapter became full.
This occurs when the fabric interconnect is configured in ENM mode and when adapter/server becomes part of more than 32 VLANs in the trunk mode. If you enter the show ip igmp snooping command, the output shows that a single veth interface is part of more groups than what the QLogic or Emulex CNA can handle.
Symptom When a DCBXP peer on a physical interface sends two different unique identifiers in the protocol data unit in the same session, a DCBXP process error disables the port.
Workaround Enter the following commands and then wait for a minute for the port to come up:scope chassis chassis-idscope server server-idscope adapter adapter-idscope ext-eth-if ext-eth-if-idset adminstate reset-connectivitycommit-buffer
Symptom When a blade server is powered on with a faulty DIMM, the BIOS POST detects the faulty DIMM and two SEL entries are sent to the BMC. One entry is logged for each DIMM.
Workaround Enter the BIOS setup and navigate to the Advanced > Memory Configuration menu to distinguish the faulty DIMM from its neighbor.
Symptom A blade server cannot boot from the HDD when the following boot order is set in the service profile:
Workaround Set a different boot order. The HDD can be the first, second, or third boot device, but not in the order specified in the symptom.
Symptom Installing EFI Native SLES 11 is currently not supported in this release.
Symptom One vNIC defined in the UCS Manager service profile boot order results in two BIOS vNICs.
Workaround Avoid defining two different pxelinux.cfg/<MAC> files that have different boot/install instructions. When booted, both vNICs should execute the same PXE configuration.
Symptom If the LAN Link (NetBoot) is the only boot device in the boot order and it fails for any reason, the local disk is selected.
Workaround Ensure that at least two bootable devices are present in the boot order. If the NetBoot fails, the other device in the boot order is selected.
Symptom Upgrading from an earlier software version may result in a falsely-reported bad software image on the BMC controller.
Workaround Remove and reinsert the blade server to cold boot the BMS and clear the erroneous report.
Symptom Without pin-group configuration, server interfaces are dynamically pinned to fabric interconnect uplink interfaces, and the pinning is redistributed as uplink interfaces go up or down. In some situations, the distribution of server interfaces across uplink interfaces is not even.
Symptom When a fabric interconnect boots, the "The startup-config won't be used until the next reboot" message appears on the console. Fabric interconnect configuration is controlled by the UCS Manager, so this message has no meaning on the fabric interconnect configuration and has no functional impact.
Symptom Console logon user names on the fabric interconnect are not case sensitive. For example, there is no differentiation between admin and ADMIN.
Workaround Use case insensitive user names.
Symptom When you set TFTP core exporter IP address through the UCS Manager, the fabric interconnect accepts the address even if it is malformed; for example, 188.8.131.522. This has no functional impact to core exporter with valid IP addresses.
Workaround Remove the malformed core exporter IP address.
Symptom CDP packets are not being sent by the fabric interconnect on virtual interfaces. Hosts will not be able to obtain CDP neighbor information from the fabric interconnect.
Symptom When the system is under high stress, with repeated port flapping (ports rapidly going up and down) and default (native) VLAN change, the FWM process may core and cause the fabric interconnect to reload.
Symptom Changing the QoS class of service value for FCoE traffic can disruptive FCoE traffic. FC I/O would be dropped.
Workaround The QoS architecture should be planned before deploying. Be aware that a change in class of service value for FCoE traffic can disrupt FCoE traffic.
Symptom The show cdp neighbor CLI command does not display information for CDP neighbors seen from the management interface, nor does it display the fabric interconnect CDP information corresponding to the management interface.
Symptom The fabric interconnect management interface IP address does not get propagated to the upstream MDS Fibre Channel switch. As a result, the Fabric Manager is not able to discover the fabric interconnect, and is not able to retrieve any information about the fabric interconnect via SNMP.
Workaround As a result, the Fabric Manager cannot be used with UCS for such things as zoning and VSAN management, Use the upstream MDS Fibre Channel switch CLI for zoning and VSAN management.
Symptom Under some circumstances, syslog messages are not shown on a terminal monitor even though logging on the terminal monitor is enabled globally and on a particular session.
Workaround Use a different syslog destination, such as console, remote server, or local log file.
Symptom When multiple SNMP trap hosts are configured, SNMP traps are only sent to the first host.
Workaround Configure only one SNMP trap host.
Symptom On rare occasions, the link between the UCS 6100 fabric interconnect and the UCS 2100 fabric extender does not come up because of a "FEX-fabric sfp invalid" error.
Workaround Perform the following steps to reset the server port:
1. From the UCS Manager GUI, click the Equipment tab in the Navigation pane,
2. In the Equipment tab, expand Equipment > Fabric Interconnects and click the fabric interconnect with the port to be reset.
3. In the Physical Display section of the Work pane, right-click the server port in the "FEX-fabric sfp invalid" state and choose Disable.
4. Right-click the same server port and choose Enable.
5. If the link does not recover, replace the SFP transceiver on the UCS 2100 fabric extender side of the link.
Faults and Alerts
Symptom Even though there are no POST failures in the chassis, the UCS Manager sometimes displays the overall status of the chassis as POST-failure.
Workaround Check the chassis POST messages for errors, and if there are none, ignore the UCS Manager status message.
Symptom In rare cases the UCS Manager reports the link absence fault between the fabric interconnect server port and the fabric extender during the internal inventory collection. The following is an example of such a fault:***************************Severity: ClearedCode: F0367Last Transition Time: 2009-07-15T11:47:49ID: 646445Status: NoneDescription: No link between fabric extender port 2/1/1 and switch A:1/9Affected Object: sys/chassis-2/slot-1/fabric/port-1Name: Ether Switch Intfio Satellite Connection Absent Cause: Satellite Connection AbsentType: ConnectivityAcknowledged: NoOccurences: 1Creation Time: 2009-07-15T11:46:49Original Severity: MajorPrevious Severity: MajorHighest Severity: Major*******************************
Workaround Ignore the fault message; it will automatically get cleared after one minute. This will not impact the data path.
Symptom On a system with five or more chassis, the following sequence of events causes the system to not be HA ready for up to five minutes:
1. Discover all chassis
2. Wait for HA READY
3. Decommission all chassis
4. Recommission all chassis
Workaround Wait for HA READY.
Symptom After a full restore of the primary fabric interconnect, the subordinate fabric interconnect installation may temporarily fail with the following message:Enter the admin password of the peer switch:Connecting to peer switch... unable to connect! Password could be wrong.Hit enter to try again or type 'restart' to start setup from beginning...
Workaround Retry the fabric interconnect installation.
Symptom When the HTTP or HTTPS service is set to a port that is already in use by another service, the web process fails to restart.
Workaround Check the ports that are in use by other services and assign a port that is not being used to the HTTP or HTTPS service.
Symptom Hardware revision numbers for fabric interconnect components are not populated in the UCS Manager.
Workaround Perform the following steps to determine the revision number for a fabric interconnect component:
1. Enter the connect nxos command to connect to the native NX-OS CLI.
2. Enter the appropriate show sprom component command and look for H/W Version: field in the command output.
Symptom All DIMMs are reported to be in array 1, and maximum memory for a blade is reported as 192 GB.
Workaround DIMMs may be physically located in array 1 or array 2, and the maximum memory is 96 GB. This is a display-only issue, and does not affect functionality.
Symptom When you create, delete, or modify the DNS, domain name, or time zone configurations, the operation is not listed in the audit log records.
Symptom The average, maximum and minimum statistics values are displaying 0.
Workaround These are actually the average, maximum and minimum change in values between collection intervals. If there is no change in values between collection intervals, then 0 is the correct value.
Symptom When the RHEL 5.3 XEN kernel is booted, the "MCFG area at a0000000 is not E820 reserved" error message appears as a part of the boot messages.
Workaround This is a known issue and a misleading error statement (to be removed by Red Hat in the 5.5 release). It does not affect any functionality of the system. Ignore the error message.
Symptom In the UCS Manager GUI, you can only select port channels and individual uplink ports as pin targets, but the UCS Manager CLI allows you to also select port channel member ports as pin targets.
Workaround Port channel member ports should not be selected as pin targets, even if the UCS Manager CLI allows you to.
Symptom An attempt to configure hard pinning on a vHBA after a service profile is associated with a server may fail.
Workaround Perform the following steps:
1. From the UCS Manager CLI, enter the connect nxos command.
2. From the NX-OS CLI, enter the show npv flogi-table command.
3. Determine which uplink Fibre Channel port the FC VIF corresponding to the vHBA is operationally pinned to. If it is pinned to a different uplink Fibre Channel port compared to where you want it pinned, the hard pinning configuration fails.
4. Hard pin your vHBA to the same uplink Fibre Channel port where it is operationally pinned, then you can successfully hard pin to a different uplink Fibre Channel port.
Pools and Policies
Symptom The assignment of servers to pools in a dynamic pool are not dynamically assigned. To have a server assigned to a pool, it must be re-acknowledged. This should happen automatically based on the server pool policy qualifications as soon as the qualification is associated to a pool.
Workaround Re-acknowledge the server to enable it as a candidate for pool assignment.
Symptom Creation of a service profile by a server autoconfiguration or server inheritance policy does not delete service profiles created previously by a policy. For example, if a server autoconfiguration policy is triggered by a server, the server is removed and a different server autoconfiguration policy is triggered. Each of the server autoconfiguration policies would have created service profiles. (The service profile would be created when discovery completes or server re-acknowledged.)
Workaround Delete the previously auto-created service profiles; You only need to do this when a new policy creates service profiles for the same server, either because a new server was inserted into the slot, or rules changed and the server was re-acknowledged.
Symptom The vNIC templates are not exported when you backup all system and logical configuration settings (the All Configuration option).
Workaround Create the vNIC templates after you import the configuration.
Symptom When defining a statistics threshold in a given direction (above normal or below normal), if the current update is in the opposite direction, an invalid fault for the opposite direction is raised.
Workaround Use 0 as the normal value when defining above-normal thresholds so that an opposite direction threshold evaluation will not occur. Use a large normal value when defining below-normal thresholds.
Symptom User-defined statistics threshold policies that are set at the vNIC or vHBA level are not evaluated for vNIC and port statistics. Within a vNIC or vHBA configuration, you can specify a statistics threshold policy that is used for evaluating threshold violations, but this is not working correctly.
Workaround Define vNIC or vHBA level thresholds within the default statistics threshold policy under root organization.
Power and Thermal
Symptom The Admin and Oper server power states may become out of sync when the server crashes, the front power button is used, or you shut down the server from the OS.
Workaround Use only the UCS Manager to change the server power state.
Symptom The I/O module (IOM) temperature sensor statistics have incorrect values. This impacts only the statistics display. There is no impact on generating faults when the sensor temperature crosses limit thresholds.
Symptom Do not mix local disk and SAN boot targets at same time.
Workaround Define one boot option at a time under the Storage Type.
Symptom Local disk removal and insertion is not detected.
Workaround Select the Re-Acknowledge Server option in UCS Manager GUI to discover the server.
Symptom The disk scrub policy needs enhancements to meet DOD compliance.
Symptom Removing a vHBA from the storage boot order in an updating service profile template does not update the boot order in service profile instances based on the template.
Workaround Manually update the boot order in the service profile instance.
Symptom If a service profile is created through a server autoconfiguration policy and from a template associated with a server pool, changes to the template may not get applied on the server. This applies only to service profiles that were automatically created through a server autoconfiguration policy.
Workaround Perform the following steps:
1. If a server autoconfiguration policy is used, do not associate it with a template that uses server pool.
2. Acknowledge the server to manually apply the template changes and reboot the server.
Symptom SNMP shows the fabric inteconnect name rather than system name.
Symptom An SNMP user name can not be the same as a local user name.
Workaround Select an SNMP user name that does not match any local user name.
Symptom Any SMASH command entered with wrong option should give "INVALID OPTION" error message.
Symptom HDD metrics (block size, number of blocks, MB size, etc.) are not displayed in the GUI Server > Inventory > Storage tab after you hot-swap or remove and reinsert the HDD. This is because the UCS Manager can only read the HDD metrics when the PnuOS (Utility OS) boots after the HDD reinsertion.
Workaround This does not impact operation of the HDD, the rest of UCS, or UCS Manager functionality. Re-associate a service profile or re-acknowledge a server to cause the PnuOS to boot and discover the new HDD metrics.
UCS Manager CLI
Symptom Statistics counters cannot be cleared using the UCS Manager CLI.
Workaround Clear the counters using the UCS Manager GUI.
Symptom In a HA setup, when you try to connect to the subordinate fabric interconnect on an out-of-band interface, the CLI may not respond for up to five minutes to any commands you enter, and displays the "DME is not ready" error message. After the five minute delay, you can enter CLI commands.
Workaround Do not use an out-of-band interface to connect to the subordinate fabric interconnect, or wait five minutes for the "DME is not ready" error message to clear.
UCS Manager GUI
Symptom A service profile can be assigned to a server that has a service profile already associated to it. When you perform this operation In the GUI, the list of servers is divided between available and all. If you make a decision to pick a server that already has a service profile, you must be aware that the new profile will replace the old one, and the operating system will be brought down as part of the blade reboot process.
Workaround Do not assign a profile to a server that is not available unless you are sure that you want to replace the existing profile with a new one.
Symptom In some cases, if a BIOS recovery is performed several times, the General Tab of the Server Navigator GUI may have both Boot Server and Shutdown Server grayed out, even though the server is associated to a service profile.
Workaround Select a different object in the navigation tree, such as a chassis, and then select the Server again.
Symptom In the Service Profile Navigator GUI, when changing the boot policy type, such as from no boot policy to a specifically-created boot policy, clicking Reset Values fails to reset the page to the initial status.
Workaround Select a different object in the navigation tree, such as the root organization, and then select the service profile again.
Symptom Sometimes, when modifying the Dynamic vNIC Connection Policy of a service profile, the Dynamic vNIC Connection Policy drawer (in the Policies Tab of the Service Profile Navigator) does not immediately reflect the new value.
Workaround To see the new value, select a different object, such as the root organization, and then select the service profile again.
Symptom Sometimes, when modifying the Serial Over LAN SOL Policy of a service profile, the Serial Over LAN Policy drawer (in the Policies Tab of the Service Profile Navigator) does not immediately reflect the new value.
Workaround To see the new value, select a different object, such as the root organization, and then select the service profile again.
Symptom When several KVM Consoles are launched, the SUN JRE sometimes reports an error and the KVM Console fails to launch.
Workaround Launch the KVM Console again.
Symptom After changing the boot order and saving the changes, a GUI pop up appears stating that "There are uncommitted changes. Are you sure you want to close the window and discard the changes?" even though the change has been committed.
Workaround Click Yes to close the pop up. The changes you saved will be preserved.
Symptom The node-wwnn object config-state is always displayed as "not-applied" even though it is deployed on the server during association.
Workaround Ignore the "not-applied" config-state for the node-wwnn object.
Symptom In the UCS Manager GUI, if the Reboot on boot Order Change checkbox is checked for a boot policy, and if CD-ROM or Floppy is the last device in the boot order, then deleting or adding the device does not directly affect the boot order and the server does not reboot.
Symptom If a VSAN, SAN pin group, VLAN, or LAN pin group is changed, the server reboots only if the Reboot on boot Order Change checkbox is checked for the associated boot policy in the UCS Manager GUI,
Symptom Fibre Channel port and server port events do not appear on the Fibre Channel port and server port Events tabs.
Workaround Look on the Admin Events tab for Fibre Channel port and server port events.
UCS Manager XML API
Symptom Internally, timestamp properties (in faults and statistics) are represented in milliseconds. but the string value is represented in the form YYYY-MM-DDTHH:MM:SS. With the loss of precision in the string value, the XML API equality filter cannot be used for resolve queries.
Workaround Use the greater than or equal filter or the less than or equal filter instead of the equality filter.
Symptom The UCS Manager powers off the server after you change the UUID in a service profile.
Workaround Before you change the UUID, perform the following steps in the UCS Manager GUI to shut down the server:
1. In the Servers tab, expand Service Profiles, and select the service profile for which you want to change the UUID.
2. In the Work pane, click the General tab.
3. In the Actions area, click Shutdown Server.
4. If a confirmation dialog box appears, click Yes.
Symptom When the time-based UUID generation method is used in the uuidgen tool, the system reports duplicate UUIDs because UUIDs are validated based on their suffix, whereas time-stamp method creates UUIDs with unique prefixes, but non-unique suffixes.
Workaround Use the random method in the uuidgen tool to ensure that suffixes are unique.
For related documentation for the Cisco Unified Computing System refer to the Cisco UCS Documentation Roadmap.
Obtaining Documentation and Submitting a Service Request
For information on obtaining documentation, submitting a service request, and gathering additional information, see the monthly What's New in Cisco Product Documentation, which also lists all new and revised Cisco technical documentation, at:
Subscribe to the What's New in Cisco Product Documentation as a Really Simple Syndication (RSS) feed and set content to be delivered directly to your desktop using a reader application. The RSS feeds are a free service and Cisco currently supports RSS version 2.0.
This document is to be used in conjunction with the documents listed in the "Related Documentation" section.
Cisco and the Cisco Logo are trademarks of Cisco Systems, Inc. and/or its affiliates in the U.S. and other countries. A listing of Cisco's trademarks can be found at 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. (1005R)
Release Notes for Cisco UCS Manager Release 1.0(1e)
© 2009 Cisco Systems, Inc. All rights reserved.