Skip to content

Network Access Control

Being on the VPN does not mean you can reach every device on it. Access is scoped to the people who are supposed to have it, and ARROW keeps that scoping in sync with your consultant assignments so you never have to write firewall rules by hand. Assign a consultant to a device in the Console, and their VPN access to that device follows automatically. Unassign them, and it goes away.

You do not manage any of this directly, but you can review all of it. Open the VPN section and use the Policies and Groups tabs to inspect the rules and groups that back your access, or the Access Graph tab to see the result per person.

The Policies tab of the VPN section in the ARROW Console The Policies tab of the VPN section in the ARROW Console

The Policies tab, where the access rules that control device access are listed with the ones ARROW maintains marked by a lock

The Policies tab of the VPN section in the ARROW Console The Policies tab of the VPN section in the ARROW Console

When you assign a consultant to a device, ARROW creates the VPN groups and access policy that let that consultant connect to it, and the access takes effect within a few minutes. When you remove the consultant, ARROW revokes the access just as automatically. There is nothing to configure on the VPN itself; the device assignment is the control.


Access policies are intentionally one-way. Consultants can reach devices over SSH, VNC, and other management protocols, but devices can never open connections back toward consultant workstations. This matters most in exactly the situation ARROW is built for. If a device you are testing from gets compromised, it cannot turn around and pivot into your own machines or the rest of your infrastructure.


Under VPN, the Policies, Groups, and Access Graph tabs show the access model in action.

The Policies tab lists every access rule on your network with its sources, destinations, and protocol. The rules ARROW maintains for you carry a lock badge and cannot be edited here, but they are worth understanding. The users-to-infrastructure-tcp and users-to-infrastructure-udp rules let consultants reach ARROW devices, per-device rules grant access to specifically assigned consultants, and separate rules cover virtual machines. Rules your organization creates itself sit in the same list and stay fully editable (see VPN Management).

The Access Graph tab turns those rules into an answer per person. Select a consultant to see the devices they sign in from, every peer those devices can reach, and the rules and engagements behind that access.

The Access Graph tab of the VPN section in the ARROW Console The Access Graph tab of the VPN section in the ARROW Console

The Access Graph, tracing one person through the rules that allow them to the groups and engagements they reach

The Access Graph tab of the VPN section in the ARROW Console The Access Graph tab of the VPN section in the ARROW Console

The graph is also the fastest way to check that someone is enrolled correctly. A workstation that signed in through your identity provider appears under Signs in from for that person; a machine that joined with a setup key does not, because it carries no identity. If a consultant reports that they are connected but cannot reach anything, look here first (see VPN Connection Troubleshooting).

The Groups tab shows how peers are organized:

GroupDescription
usersUser workstations authenticated via IDP
pvePhysical ARROW devices (Proxmox hosts)
pvmVirtual machines on physical ARROW hardware
vmVirtual machines on other infrastructure
socksSOCKS proxy hosts

Groups maintained by ARROW carry the same lock badge as the managed rules. You can create and manage your own groups alongside them.


Access is driven entirely by device assignments, so you manage it the same way you manage a device. To grant access, open the device in the Console, edit its request or settings, add the consultant to the assignment list, and save; access appears within minutes. To revoke it, remove the consultant from that same list and save, and their access is pulled immediately.


Virtual machines follow the same model, set up for you when the VM is provisioned. ARROW creates a group for the VM, a group for its assigned consultants, and a one-way policy connecting the two. When the VM is marked complete, all of it is cleaned up automatically, including the peer registration, both groups, and the policy.


For this to work, each consultant needs an ARROW account in your organization. You do not create a separate VPN account for them, because when you add the user in the Console, ARROW gives that account VPN access at the same time. Consultants then sign in through your organization’s identity provider with the same email address they use for the Console. If a consultant is assigned but cannot connect, a mismatched email is the first thing to check.


Because access tracks assignments, keeping assignments honest is what keeps access tight. Review consultant assignments periodically, remove people promptly when they roll off a project, and note in the device request why a given consultant is assigned so the reasoning is not lost.