This chapter discusses various issues that may be encountered while configuring the examples in this guide. A problem statement is given for each scenario, followed by one or more possible solutions. For additional troubleshooting guidance, see the Dell EMC PowerEdge MX Networking Deployment Guide.
Problem | Description | Solution |
Dropped packets between VMs for 15 seconds after switch reboots | Rebooting the MX9116n FSE on the PowerEdge MX chassis while passing traffic between VMs deployed on MX740c compute sled and VMs outside of the MX chassis deployed on R730xd rack server causes 3 to 5 Requested Time Outs and dropped packets for up to 15 seconds. | This is caused by the MX9116n FSE on the PowerEdge MX chassis becoming the Spanning Tree root. To resolve, make an Upstream switch the STP Root, not the MX9116n FSE. In the topology mentioned here, whichever switch has the lower priority number increases the likelihood of the bridge to become STP Root. Run the commands detailed in the SmartFabric OS10 User Guide to make an upstream switch the STP root. See OME-M and OS10 compatibility and documentation to find the appropriate version. |
NPAR issues on a virtual switch | When enabling NPAR on QLogic 41xxx series adapters (for example QLogic 41262 in conjunction with either standard virtual switch or distributed virtual switch) on ESXi operating system, connecting multiple NIC enabled partitions from a single physical adapter port to the same virtual switch is not a supported configuration. This can cause unexpected issues on the network. | Implement a supported configuration, by attaching one NIC enabled partition per physical adapter port to any single virtual switch and utilize multiple distributed port groups in the case of a distributed virtual switch to separate network workloads. If a user requires multiple NIC enabled partition ports per physical adapter port, then separate virtual switches are needed and each virtual switch would have one NIC enabled partition port per physical adapter port |
Port flapping with LLDP advertisements enabled on vDS | When MX9116n FSEs are connected to MX7116n FEMs in an MCM Group, port flapping has been observed when a user enables LLDP (Link Layer Discovery Protocol) under Discovery and selects Advertise under Operation on the distributed virtual switch settings with an End to End ESXi scenario. Note: This issue is not seen when using CDP. | To resolve, it is recommended to not enable LLDP in the distributed virtual switch settings, as this can cause more than the expected number of LLDP neighbors (one for iDRAC and one for the adapter port) on the switch interface. To disable discovery protocol under vDS, select vDS under the Networking listing, choose Configure, then click Edit. Click the Advanced listing and from the Type drop-down, disable the Discovery Protocol or select CDP as per topology. |
Problem | Description | Solution |
FCoE session will not establish | In a scenario mentioned where MX9116n FSE is connected to PowerEdge MX740c or MX840c compute sled with QLogic 41262 mezzanine adapter, configured to use FCoE. When a connecting IOM is reset, FCoE session may not establish or re-establish. The MX9116n includes two universal QSFP28 ports, each of which can break out to 4x16Gb or 4x32Gb fibre channel interfaces, making a total of 8 Fibre Channel Forwarder (FCF) ports. This issue is most often seen in MX9116n FSE configurations with 4 or more FCF ports configured in a vFabric. In ESXi OS vmkernel.log, this issue is most easily found with repeated | To resolve this, For ESXi 6.5/6.7 OS, update qedf to 1.3.42.0 or newer as well as associated qedentv/qedrntv and qedil/qedi to versions matching the Dell EMC qualified combinations. |
Inconsistencies found in VMNIC enumerations with NPAR | In a vDS environment it is important that vmnic enumeration are consistent between hosts in order to avoid misconfiguration. In NPAR setups, there is a risk of unaligned PCI to vmnic enumerations depending on which NPAR configuration that existed the first time ESXi booted. | Make sure to configure the final NPAR configuration (with the server template) before booting ESXi for the first time. If the order of MAC-addresses is in sequence, then the order is most likely correct. Another possible solution is to reconfigure in accordance with the How VMware ESXi determines the order in which names are assigned to devices KB article. |
MX7116n FEMs are not discovered when creating a SmartFabric | In the PowerEdge MX chassis, the Link Layer Discovery Protocol (LLDP) advertisements from the Blade NICs may not be visible to the IOMs. Running the "show lldp neighbors" command from the IOM will not list the NICs. Note: For an example of the IOM not listing the NIC, see NICs not listed in show lldp neighbors output. In the Blade iDRAC, the status of the NICs are shown as Unknown, and the Switch Connection ID and Switch Port Connection ID are shown as Not Applicable. This issue may prevent MX7116n from being discovered when creating a SmartFabric. | To resolve the issue, there are few points that need to be considered. In the In end-to-end ESXi scenarios mentioned here:
|