Managing NICs

Use the command line to view and modify the physical ports associated with the OVS bridge in the virtual distributed switch, as well as the bonding modes of these ports.

This command is mainly applicable in the following scenarios:

• Replacing a NIC.
• Changing the NIC associated with the OVS bridge on a host.
• Modifying the bonding mode of the NIC associated with the OVS bridge.

Precaution

• Replacing a NIC requires modifying the virtual OVS bridge information on the node host and synchronizing the changes to the database.
• Replacing a NIC requires that no data recovery is in progress on the node. If data recovery is in progress, the time to complete it may vary depending on the workload. It is recommended to replace the NIC when the node is under low workload.

Obtaining the NIC bond name corresponding to the network bridge

Obtain the bond name corresponding to the OVS bridge on a running node.

Procedure

Run the following command on the node:

network-tool get-bond-name --ovsbr_name <ovsbr_name>

--ovsbr_name <ovsbr_name> The OVS bridge name for which to obtain the bond name.

Usage example

network-tool get-bond-name --ovsbr_name ovsbr-n5b9qlfm2

Output example

2022-05-18 12:44:33,892 [72353] [INFO] bond_name: bond-m-dd417f1a

Setting OVS bridge parameters

Use the command line to set the following parameters of an OVS bridge:

• The OVS bridge name
• The NIC associated with the OVS bridge
• The bonding mode of the virtual distributed switch (VDS) associated with the OVS bridge

Procedure

Run the following command on the node where the OVS bridge resides:

network-tool change-bridge {argument list}

Parameter description

Usage example

Configure parameters for the OVS bridge named ovsbr-mgt. Update a single network port (eth0) to dual network ports (eth0 and eth1). The associated VDS uses the balance-slb bonding mode of OVS bonding, with a rebalance interval of 80 seconds. The new bond name corresponding to the OVS bridge is bond-mgt.

network-tool change-bridge --ovsbr_name ovsbr-mgt --old_port_name eth0 --nics "eth0 eth1" --bond_mode balance-slb --rb_interval 80 --bond_name bond-mgt --mcast_snooping disable

If RDMA is enabled for the storage network, configure parameters for the OVS bridge that is originally named ovsbr-stor. Change the Linux bond 802.3ad mode to OVS bond balance-tcp mode (eth0 and eth1). The new bond name that corresponds to the OVS bridge is bond-storage.

network-tool change-bridge --ovsbr_name ovsbr-stor --old_port_name bond2 --nics "eth0 eth1" --bond_mode balance-tcp --bond_name bond-storage --mcast_snooping disable

Output note

If change bridge finish is displayed, the OVS bridge information has been updated successfully.

Synchronizing updated OVS bridge NIC information to the database

Synchronize the updated OVS bridge information to the database.

Procedure

Run the following command on the node where the OVS bridge resides:

network-tool sync-bridge {argument list}

Parameter description

Usage example

network-tool sync-bridge --ovsbr_name ovsbr-mgt --bond_mode balance-slb --bond_name bond-mgt --nics "eth0 eth1"

Output note

If sync successfully is displayed, the data has been successfully synchronized.

Precaution

After all nodes have completed OVS bridge information synchronization, if the last node that executed the network-tool sync-bridge command has a single network port while other nodes have multiple network ports, you need to select any node with multiple network ports and execute network-tool sync-bridge again.

Obtaining the physical network ports in a Linux bond

Procedure

Run the following command on the node where the OVS bridge resides to obtain information about the physical network ports contained in the Linux bond:

network-tool get-linux-bond-list

Output note

Output the physical network ports in a Linux bond.

Testing the connectivity of the OVS bonding network ports

Procedure

Run the following command on the node where the OVS bridge resides to test the connectivity of the physical network ports contained in the OVS bond.

Information:

• This command can only test the connectivity of network ports bonded in the VDS of the management network and storage network.
• During the test, the command performs ifdown/ifup operations on the bond network ports, which may cause a brief network interruption.

network-tool check-ovs-bond-nics --bond_mode <bond_mode>

Usage example

network-tool check-ovs-bond-nics --bond_mode "active-backup"

Output note

The output shows the connectivity test results of each network port in the bond to other nodes.

• If check bond nic complete: all nic connection perfect! is displayed, the connectivity of all network ports in the bond is normal.
• If ['eth2 can not ping xx.xx.xx.xx', 'eth2 can not ping xx.xx.xx.xx'] is displayed, the eth2 network port in the bond cannot reach the target nodes.

Changing the node management network IP

Procedure

Run the following command on a cluster node to change its management IP address:

network-tool change-manage-ip <ip> <netmask> <gateway> [--vlan_id <vlan_id>]

Output example

Displays the configuration process. If the output shows change management ip address success!, the management IP address of the node has been successfully modified.

Migrating the default VM network to another VDS

After cluster deployment is complete, a VM network named default is automatically created on the VDS where the management network resides. If you need to migrate the default VM network to another VDS, follow the instructions in this section. After migration, the MTU of the default VM network will be reset to 1500.

Prerequisite

• The default VM network does not have QoS configured.
• The default VM network is not associated with any virtual machines, VM snapshots, or VM templates that have virtual NICs configured.

Procedure

Run the following command on any cluster node to migrate the default VM network to another VDS:

network-tool migrate-default-network --vds_name <vds_name>

Usage example

network-tool migrate-default-vm --vds_name vds_new

Output note

• If the migration fails, the corresponding failure reason is displayed.
• If the migration succeeds, the message migrate default vm network to vds_new success! is displayed.