WEKA tenant clusters use VLAN tagging to enable isolated network communication between clusters and their clients.
WEKA support for IEEE 802.1Q VLAN encapsulation ("tagged VLAN IDs" or "tagged VLANs") enables isolation and segregation of network traffic while still granting connectivity between WEKA clients and backend servers.
A WEKA tenant cluster operates on a single VLAN ID, where all containers within that cluster must use the same VLAN ID. All clients connecting to a tenant cluster must use that cluster's assigned VLAN ID. When using the WEKA Kubernetes Operator, this VLAN ID consistency between the tenant cluster and its clients is automatically maintained.
In multi-cluster environments, each tenant cluster can operate on a different VLAN ID. For example, you can assign VLAN ID 100 to Tenant Cluster A and VLAN ID 200 to Tenant Cluster B, providing network isolation between clusters.
To enable WEKA tagged VLAN support, add the desired VLAN IDs to the switch ports connected to WEKA backends. It is common to include one untagged VLAN and multiple tagged VLAN IDs.
After configuring the switch, update the Linux system interfaces to recognize the VLAN IDs and verify connectivity.
Use the following procedure on each WEKA backend and each WEKA stateful client. You do not need to perform this procedure on WEKA stateless clients because it is automatically applied during the mount command.
Procedure
Associate the VLAN with the network interface using one of the following methods based on your naming preference:
Assign a tag to a specific device:
weka local resources net add <nic> --vlan <tag>
Example:
weka local resources net add mlnx0 --vlan 501
Add a dedicated VLAN interface where the tag is inferred from the name:
weka local resources net add vlan<tag>
Example:
weka local resources net add vlan501
Restart all containers on the server to apply the network configuration updates.
To determine which tagged VLANs are configured, query the local resources of a container using a command like this:
Example:
You can also display the cluster container network data with this command by requesting verbose output with -v to list the VLAN ID or by using syntax like -o id,hostname,name,ips,vlan:
When running a mount command on a WEKA stateless client where the backends have a tagged VLAN, specify the same VLAN in the mount command as follows:
Mount a filesystem with a specified NIC and VLAN tag:
Example:
Include gateway, IP, and netmask (in CIDR format) for advanced configurations:
Example:
Syntax guidelines:
Include additional named parameters (for example, gw@, ip@, netmask@) directly in the command syntax.
Alternatively, use the legacy style by specifying name-value pairs after the positional parameters.
Separate all parameters using /.
This syntax is also supported for the weka local setup container --net ... command.
Last updated
weka local resources -C <containername>weka local resources -C drives0weka cluster container net -vmount -o net=<nic>/vlan@<tag> <backendip/filesystem> <mountpoint>mount -o net=mlnx0/vlan@501 10.10.0.10/default /mnt/wekamount -o net=<nic>/vlan@<tag>/gw@<gateway>/ip@<ip>/netmask@<netmask> <backendip/filesystem> <mountpoint>mount -o net=mlnx0/vlan@501/gw@192.168.1.1/ip@192.168.1.10/netmask@22 10.10.0.10/default /mnt/weka