General Settings
Overview
Settings → General holds the controller-wide parameters of a Netris Controller. These settings apply to every Site, VPC, and device the Controller manages; there is no per-Site override for any of them. Most deployments set the Controller FQDN and Local Repository during installation and leave the rest at defaults.
To change a setting, navigate to Settings → General and click Edit. Changes take effect immediately on save; no Controller restart is required.
Settings
Setting |
Default |
Description |
|---|---|---|
Netris Version |
— |
Read-only. Version of the Netris Controller software. |
Netris Auth Token |
generated at install |
Shared secret that Netris agents on switches, SoftGate nodes, and DPUs present when registering with the Controller. The agent installation one-liner embeds it. Use the eye icon to reveal it and the copy icon to copy it. |
Noreply email address |
Sender address for email the Controller generates (alerts, notifications). Set it to an address your mail system accepts. |
|
Controller FQDN or IP address |
— |
Address at which switches, SoftGate nodes, DPUs, and users reach the Controller. Used in the agent installation one-liner and in generated links. Use an FQDN when the Controller is secured with a certificate issued for that name. |
Approval check for source address matching |
true |
When enabled, an agent’s registration request is approved only if it arrives from the IP address recorded for that device in Inventory. |
Optimized deployment of ACLs for subinterfaces |
true |
When enabled, Netris installs ACL entries once per parent interface instead of once per subinterface where the result is equivalent, reducing rule count on the switch. |
L4LB: Number of probes in each health check iteration |
5 |
Probes sent to each L4 Load Balancer backend per health check iteration. |
L4LB: Number of failures to mark a backend as failed |
3 |
Consecutive failed iterations before a backend is removed from the pool. |
L4LB: Number of consecutive successes for the probe to be considered successful after having failed |
1 |
Consecutive successful iterations before a failed backend is returned to the pool. |
L4LB: Interval (seconds) between health checks |
10 |
Time between health check iterations. |
K8s Layer 4 load-balancer timeout (ms) |
2000 |
Timeout for L4 Load Balancer services created through the Netris Kubernetes integration. |
System ASN range |
4200000000 – 4209999999 |
Pool from which Netris assigns a BGP Autonomous System Number to each switch whose AS Number is set to assign automatically in Inventory. Site Public ASNs and E-BGP peer ASNs must not fall inside this range. See Netris Site. |
Local Repository |
disabled |
When enabled, agents download the Netris agent installer, packages, and NOS images from the repository at this URL instead of the Netris public repository. Required for air-gapped deployments and for ZTP. See Installing HA Netris Controller in Air-Gapped Environments and Zero Touch Provisioning (ZTP). |
Controller Management Address |
disabled |
Fallback address used to reach the Controller from the management network. Netris resolves the Controller’s address for each management subnet automatically and uses that address. If no management subnet attached to the device has a resolved address, Netris falls back to this global setting when one is configured. ZTP and the agent installation one-liner both follow that order, so setting this is not required. Applies to switches and SoftGates. See Zero Touch Provisioning. |
UFM Settings
These settings apply only when the NVIDIA UFM integration is in use. See NVIDIA UFM (InfiniBand) Integration.
Setting |
Default |
Description |
|---|---|---|
Dry Run |
false |
When enabled, Netris computes the InfiniBand PKey changes it would apply to UFM and logs them without applying them. |
Debug |
false |
Enables verbose logging for the UFM integration. |
Reconcile Interval (seconds) |
30 |
How often Netris compares its intended PKey state with UFM and corrects drift. |