Network Requirements
This section describes the ports that must be open on each NextGen Gateways node. These open ports allow communication between managed resources, the NextGen Gateways, and the OpsRamp cloud.
1. Enable Communication (Mandatory)
2. Inbound Connectivity
| Name | Description | Protocol | Port |
|---|---|---|---|
| SSH | To access NextGen Gateways and agent consoles on the node from OpsRamp cloud. | TCP | 22 |
| Agents | (Optional). Accept connections from agents if agents are configured to connect to the OpsRamp cloud using the NextGen Gateways embedded squid proxy. | TCP | 3128 |
| SNMP Traps | SNMP traps from managed devices are sent to the NextGen Gateway IP address—typically the node IP in single-node deployments. In High Availability (HA) setups, use the Load Balancer IP address instead. | UDP | 162 | Syslog messages | Syslog messages from managed devices are sent to the NextGen Gateway IP address—typically the node IP in single-node deployments. In High Availability (HA) setups, use the Load Balancer IP address instead. | TCP/UDP | 514 |
Understanding on SNMP Traps, Syslogs, and Squid Proxy Inbound IP Addresses
The inbound IP address for services such as SNMP traps, Syslogs, and the Squid proxy depends on the deployment architecture of the Nextgen Gateway and the underlying infrastructure setup:
1. OpsRamp-Provided ISO/OVA Deployments
- Single-node cluster (default setup):
- All inbound traffic for SNMP traps, Syslogs, and the Squid proxy listens directly on the node’s IP address.
- Multi-node cluster:
- A dedicated load balancer IP address is automatically allocated:
- One shared IP for SNMP traps and Syslogs
- A separate IP for the Squid proxy
- A dedicated load balancer IP address is automatically allocated:
This ensures traffic is properly routed to the respective services across the cluster.
2. Custom or Bring-Your-Own Kubernetes (BYOK) Clusters
If you are using your own Kubernetes cluster instead of the OpsRamp-provided ISO/OVA:
- The inbound IP allocation depends entirely on how you’ve configured the Kubernetes Services, particularly:
- Whether you’re using LoadBalancer, NodePort, or Ingress resources.
- How your external load balancer (e.g., MetalLB, cloud-native LB) is set up to route traffic.
3. Managed Kubernetes Clusters (e.g., EKS, AKS, GKE)
In managed Kubernetes environments:
- The Service type
LoadBalanceris typically used for SNMP, Syslogs, and Squid proxy services - IP addresses are automatically provisioned by the cloud provider’s load balancer infrastructure
- The actual public or private IP address depends on:
- The Kubernetes Service configuration
- The networking mode (e.g., VPC-native, public vs private subnet)
- Any internal or external LoadBalancerClass defined in your cloud setup
⚠️ It is essential to ensure firewall and security group rules allow ingress on the expected ports for SNMP Traps (usually UDP 162), Syslogs (UDP/TCP 514), and Squid (default HTTP 3128) depending on your use case.
3. Outbound Connectivity
| Name | Description | Protocol | Port |
|---|---|---|---|
| vProbe | Connect to the OpsRamp cloud platform using the public IP addresses. | TCP | 443 |
| DNS | To resolve address *.api.opsramp.com. | TCP | 53 |
| NTP | Network time protocol is used for clock synchronization using the public IP addresses. | UDP | 123 |
Understanding on Outbound Traffic from the NextGen Gateway
Outbound traffic from the Nextgen Gateway refers to all communication initiated by the gateway toward customer-managed infrastructure and devices. This includes - but is not limited to:
- Discovery of network devices, servers, and applications
- Monitoring data collection (e.g., SNMP polling, WMI queries, API calls, etc.)
- Performance metric collection and status polling
Essentially, this is how the gateway reaches out to pull data from customer-managed assets.
IP Behavior in Multi-Node Cluster
In a multi-node Kubernetes cluster, the gateway pod (which performs discovery and monitoring) is scheduled dynamically across available nodes. This has direct implications for outbound traffic:
- Outbound traffic from the gateway will originate from the node’s IP where the gateway pod is currently running.
- In case of:
- Pod restart
- Node failure
- Cluster rescheduling
The gateway pod may move to a different node, changing the source IP for outbound connections.
OpsRamp Recommendation
To ensure uninterrupted data collection and monitoring:
OpsRamp strongly recommends whitelisting all node IP addresses in the Kubernetes cluster within the customer’s:
- Perimeter firewall
- Internal firewalls (if any)
- End device access control lists (ACLs), security groups, etc.
This ensures that regardless of where the gateway pod is scheduled, it can consistently initiate outbound connections to customer infrastructure without being blocked.
Additional Considerations
- In cloud-based or managed Kubernetes clusters, source IP behavior may depend on:
- Network plugins (e.g., Calico, Cilium, AWS VPC CNI)
- Whether NAT gateways, egress proxies, or egress load balancers are in use
- If outbound NAT or proxying is in place, you may need to:
- Whitelist the egress IP or NAT gateway IP instead of each node
- Ensure source IP preservation if devices enforce strict ACLs
- For environments with network segmentation or zero trust, ensure:
- Routing and firewall rules allow each node to access the required device subnets and ports
- DNS resolution is functional from all nodes (if hostname-based targets are used)
Summary
| Deployment Type | Outbound Traffic Source | Whitelisting Required |
|---|---|---|
| Single Node Cluster | Node IP (static) | Whitelist the node IP |
| Multi-Node Cluster | Any Node IP in the cluster | Whitelist all node IPs |
| Managed K8s (EKS, GKE, AKS) | Depends on CNI and NAT setup | Whitelist NAT IP / All worker node IPs |
Inbound Rules for k3 Nodes
Note
If you are using the OpsRamp-provided ISO/OVA, ensure the following ports are whitelisted. If you are not using ISO/OVA, you can ignore this section.| Protocol | Port | Source | Destination | Description |
|---|---|---|---|---|
| TCP | 2379-2380 | All nodes | All nodes | Required only for high availability (HA) deployments using embedded etcd. |
| TCP | 6443 | All nodes | All nodes | K3s supervisor and Kubernetes API server. |
| UDP | 8472 | All nodes | All nodes | Required only for Flannel VXLAN networking. |
| TCP | 10250 | All nodes | All nodes | Kubelet metrics collection and monitoring. |
Server Requirements
This section describes the hardware requirements to monitor the resources as per different capacity categories.
Capacity (Node-Level Configuration)
| Number of Managed Resources | Required Rerver Capacity |
|---|---|
| Up to 100 resources | 4 CPU cores, 8 GB RAM / 60 GB Disk / 1 NIC |
| Up to 500 resources | 8 CPU cores, 16 GB RAM / 60 GB Disk / 1 NIC |
| Greater than 500 resources at a single site | Deploy multiple Gateways |
Note
Capacity values are approximate and can vary depending on various factors, such as template frequency, resource type, network protocol, and the type of metrics.Default Memory Limits for OpsRamp Containers
By default, if you do not customize memory limits during deployment, the OpsRamp Bootstrap Tool applies predefined memory limits to each container, regardless of the actual resource capacity of the host node. These default limits are:
- vProbe: 4 GB
- Postgres: 1 GB
- Native Bridge: 500 MB
- Squid: 500 MB
These defaults are designed to ensure baseline performance and stability but may need adjustment depending on your environment and available resources. If you wish to customize memory limits during the registration process, please refer to the Registering NextGen Gateway Document.
Container Memory Allocation Guidelines
| Allocated Node RAM | vprobe | Postgres | Nativebridge | Squid | Redis + OS + Other System Overhead |
|---|---|---|---|---|---|
| 8 GB | 4 GB | 1 GB | 500 MB | 500 MB | ~2 GB |
| 16 GB | 8–10 GB | 2 GB | 1 GB | 1 GB | ~2–4 GB |
Tip
- When 16 GB RAM is available, vprobe can be allocated 8 to 10 GB depending on workload and Redis/system overhead.
- The remaining memory after allocating to main containers is reserved for:
- Redis POD
- OS-level services
- Logging agents and auxiliary pods
Additional Considerations
If Synthetics or Network Performance Monitoring (NPM) is enabled on this gateway, additional resources should be provisioned based on the specific requirements of those features.