Guidelines and Limitations
-
You must apply patch for any changes that happen on switch side (Nexus 3000 and/or Nexus 9000), to enable Cisco DCNM to support those features. To apply that patch to your Cisco DCNM Native HA setup, follow the steps below:
-
Stop the services on the Active node using the /etc/init.d/FMServer stop command.
-
Run patch.sh on the Active node.
-
Run patch.sh on Standby node.
Note
Services are not stopped on Standby node.
-
Start services on the Active node using the /etc/init.d/FMServer start command.
-
Stop the services on Active node using the /etc/init.d/FMServer stop command and roll back the patch.
-
Roll back the patch on the Standby node.
-
Start services on the Active node using /etc/init.d/FMServer start command.
-
-
Ensure that you've installed Visual C++ Redistributable Packages for Visual Studio 2013 64 bit before installing or upgrading to Cisco DCNM Release 11.4(1).
-
To check the status of the running Postgres database in Native HA setup, use pg_ctl command. Do not use the systemctl command.
-
Do not begin the password with Hash (#) symbol. Cisco DCNM considers the password as an encrypted text if it begins with # symbol.
-
POAP Dynamic Breakout—From Cisco NX-OS Release 7.0(3)I4(1), POAP dynamically breaks out ports to detect a DHCP server behind one of the broken-out ports. Previously, the DHCP server that is used for POAP was directly connected to a normal cable as the breakout cables were not supported. POAP determines which breakout map (for example, 10gx4, 50gx2, 25gx4, or 10gx2) brings up the link that is connected to the DHCP server. If breakout is not supported on any of the ports, POAP skips the dynamic breakout process. After the breakout loop completes, POAP proceeds with the DHCP discovery phase as normal.
Cisco DCNM leverages the dynamic breakout to simplify the fabric setup by retaining successful breakout configuration. Since dynamic breakout requires the other side of the link to be active, there are circumstances where you must manually breakout interfaces, or may notice breakout in places which are not desired. In those situations, you must adjust the ports on the Interfaces page before performing Save and Deploy in the Fabric Builder.
-
Before using the licensed features, install a Cisco DCNM license for each Nexus-managed or MDS-managed platform. For information about licensing, see the Cisco DCNM Licensing Guide, Release 11.x.
-
Depending on how a switch handles the cdp enable CLI command (enabled or disabled by default), Cisco DCNM shows this as config difference, although the Save and Deploy operation is performed to correct it. This depends on the default behavior of the switch image (that is, whether the show running-config shows the CLI or not). To address this issue, the respective policy template that is applied on the interfaces must be updated, so that the CLI is ignored during the configuration compliance check.
-
Create a free-form configuration on all the white box switches that are managed by Cisco DCNM as shown below, and deploy them on all the switches before the final Save and Deploy operation.
line console speed 115200 stopbits 2
This is only applicable to the Cisco DCNM LAN Fabric mode.
-
On Microsoft Windows 2016 Standard server, run the Cisco DCNM installation EXE file as an administrator. Cisco DCNM installation will not start on Microsoft Windows 2016 Standard server unless you set the EXE file as an administrator. To start the installation EXE file, you can right-click on the EXE file, and choose Run as administrator.
-
When the Cisco Nexus 9000v Virtual Switches are cloned, they may use the same serial number. Since Cisco DCNM discovers them using the same serial number, the device discovery operation fails.
-
Addition of FEX or breakout of interfaces is not supported in External Fabrics.
-
From Release 11.2(1), you can configure IPv6 address for Network Management for compute clusters. However, DCNM does not support IPv6 address for containers and must connect to DCNM using IPv4 address only.
-
You cannot access the Cisco DCNM Web UI, when the user system is configured with the same IP address range as that of internal subnet used by the Application Framework in DCNM. For more information, see Cisco DCNM Troubleshooting Guide.
-
The VXLAN OAM feature in Cisco DCNM is only supported on a single fabric or site.
-
You cannot configure ICAM on the Cisco Nexus 9000 Series Switches Release 7.0(3)I7(6), and therefore, the telemetry will fail until the switch issue is resolved.
-
Though you can delete PMN hosts, we recommended that you use this option with extreme caution, understanding that manual effort is needed to bring the solution back in sync.
-
Cisco DCNM in Media Controller Deployment Release 11.x does not support non-default VRFs for Cisco Nexus 9000 Release 9.3(x).
-
From Cisco DCNM Release 11.2(1), the Device Connector allows you to change the access mode via the Web UI at Administration > DCNM Server > Device Connector > Settings > General. The Cisco Intersight will not configure its device connector, and therefore, the Read-Only and Allow Control access mode in the Device Connector are not operational.
-
Cisco DCNM does not support hot snapshots. While taking snapshots, we recommend that you power off the VM. Otherwise, ensure that you uncheck the Snapshot the virtual machine's memory option.
-
Cisco DCNM does not support suspending or unsuspending of the VMs.
-
Do not install NIR on standalone DCNM
-
If NIR was installed and stopped, it does not stop service containers running on DCNM compute nodes.
If the NIR application is deleted from DCNM, a few service containers continue to run DCNM compute nodes and must be stopped manually using afw service commands.
-
When NIR/NIA applications is enabled at higher scale, that is, with 250 switches and 10000 Hardware telemetry flows, DCNM Computes nodes must be connected on all eth0, eth1, and eth2 interfaces using a 10Gig link.
-
When DCNM Tracker is enabled, the NIR LAN Telemetry feature in Managed mode and the EPL feature with the Configure my Fabric option selected, will not work. As a workaround, disable the DCNM tracker on the switches that are configured during the EPL or NIR LAN Telemetry configuration. For EPL, disable the DCNM tracker on the Spines/Route Reflectors (both RR1 and RR2). For NIR LAN Telemetry, disable the DCNM tracker on all the switches selected for telemetry configuration.
-
The DCNM installer creates a _deviceImage-0.iso in the DCNM VM folder and mounts the ISO permanently to the VM. If this ISO is removed or the CD/DVD is disconnected, the VM will not boot. The VM will enter Emergency Mode and prompt you with the message:Give root password for maintenance. If the VM is down, CD/DVD drive can be disconnected. However, after you power it up again, the VM will enter Emergency Mode and provide a prompt.
-
For leaf-leaf ports in non-VPC cases, DCNM will always push the shutdown command. If you want to bring up the port, add the no cdp enable command to the interface freeform policy on one of the ports.
For leaf-leaf or border-border connected ports in non-VPC cases, DCNM will always push the shutdown command to avoid the potential of loops in a VXLAN EVPN fabric. To bring up the port, add no cdp enable command to the interface freeform policy on one of the ports. Consequently, the link will however not be discovered and consequently not show up in the topology but the interfaces will still be up.
-
Two-factor authentication is not supported in DCNM.
-
After the eth0 IP address (for standalone deployment) or the vip0 IP address (for Native HA deployment) is modified using the appmgr update network-properties command, on the Web UI > Administration > MultiSite Manager does not display the correct IP address for AMQP.
-
When a Nexus Dashboard server is adding a Site from DCNM 11.5(1), it must reach the DCNM server over the Data Network. DCNM Data Network connectivity is defined to be over eth2 interface of the DCNM server; also known as Inband Connectivity interface in DCNM. When the eth2 connectivity of the DCNM with the Data Network Connectivity of the Nexus Dashboard is spanning multiple subnets, that is, when they are Layer3 Route connected, you must add routes in DCNM before adding the Site on ND.
To add route over the Inband Network in DCNM, on the Cisco DCNM Web UI, choose Administration > Customzation > Network Preferences. Enter the Routes to the ND Data Network over the In-band(eth2) inputs of the dashlet. For more information, see Network Preferences-Routes.
-
From Release 11.4(1), Cisco DCNM does not support syncing fabric with switches in VTP server mode. For more information, refer to CSCvx86976.
-
In Cisco DCNM SAN deployment, if the DCNM server streaming the SAN analytics is over-utilized, the Elasticsearch database service goes down. This results in performance issues. The Pipeline service may be consuming all the CPU and system resources on the Cisco DCNM server. To troubleshoot this, do the following task:
-
Stop the Pipeline service.
-
Reduce the streaming load from the MDS fabric.
-
Start Elasticsearch service.
-
Start the Pipeline service.
-
-
From Cisco DCNM Release 11.5(2), VLAN range is extended. After patch update for LAN Fabric deployment, you can set VLAN range to 4094.
-
In Cisco DCNM SAN deployment, when you enable or disable alarms on a Primary node, it will not be applied to all the nodes in the Federation. You must manually enable or disable alarms on all nodes on all servers in the Federation setup. You must restart the DCNM Server to apply the changes.
-
In Cisco DCNM SAN deployment, when you add or delete alarm policies on a Primary node, it will not be applied to all the nodes in the Federation. You must restart all the DCNM servers to apply this change on all servers in the Federation setup.
-
In Cisco DCNM SAN deployment, when you modify the server properties on Cisco DCNM Web UI > Administration > DCNM Server > Server Properties on a Primary node, it will not be applied to all the nodes in the Federation. You must manually make the changes to the server properties on all nodes on all servers in the Federation setup. You must restart the DCNM Server to apply the changes.
-
SAN Insights is not recommended on Windows Deployments, and is no longer supported from Release 11.3(1).
-
SAN Insights is best supported on Linux from Release 11.0(1), and on Cisco DCNM OVA/ISO deployments from Release 11.3(1).
-
From Cisco DCNM Release 11.3(1), you cannot download the SAN Client package from the Software Downloads page. You must install Cisco DCNM, launch Web UI to download the SAN Client and Device Manager. For more information, Cisco DCNM Installation and Upgrade Guide for SAN Deployment.
-
In Releases prior to 11.4, if you have installed a preview feature, perform the following before you upgrade to Release 11.4(1):
-
Remove the configuration from older release setup.
-
Reset the property to enable the preview feature. On the Cisco DCNM Web UI, choose Administration > DCNM Server > Server Properties. Reset the enable preview feature property.
-
-
We recommend that you do not upgrade any underlying third-party software separately. All the necessary software components will be DCNM upgrade will cause performance issues.
-
You can choose to discard the old performance manager (PM) data and continue to upgrade to DCNM Release 11.4(1). For instructions about how to drop performance manager data, see Dropping Performance Manager Data in the Cisco DCNM Installation and Upgrade guide for your deployment. If you choose to retain the old PM data while you upgrade to Release 11.4(1), we recommend that you contact Cisco TAC for further assistance.
Certain commands must not be executed on Cisco DCNM, as they may harm the functionality of various components on the network. The following table shows the commands and specifies the reason why they must not be executed.
Command |
Reason |
---|---|
systemctl restart network |
This is a common Linux command that the network administrators use when editing the interface properties. The command has shown to render the DCNM useless when converting to the cluster mode. |
ifconfig ethx y.y.y.y/zz |
Any change in the IP addresses of the DCNM nodes must be done with the appmgr update network-properties command. This includes changing the FQDN, adding static routes, adding/removing NTP servers etc. |