Rules
Define who can reach what, over which protocol and port, with rules — and learn how rules are applied.
In nixscale, no traffic is allowed by default. Two devices on the same network can't reach each other unless a rule allows it. A rule defines one permission: source → destination → protocol and port.
How rules work
- The source is the devices that can open a connection; the destination is the devices they connect to.
- The source or destination can be a people group (the personal computers of the people in it) or a label (the machines that carry it).
- A rule is one-way: the source can connect to the destination, and replies to an allowed connection come back automatically. For the destination to open its own connections to the source, you need a separate rule in the other direction.
- Only the protocols and ports you choose in the rule are opened.
Example: developers group → db label → TCP 5432. Developers' computers can connect to the database servers on the PostgreSQL port; the servers can't open connections to developers' computers.
Add a rule
Open the form
From the network page's menu, go to Rules and click Add rule.
Choose source and destination
In the Source and Destination lists, groups are under 👤 People groups and labels under 🖥 Machine labels. You can create a missing label right here with + New label.
Choose protocol and port
Pick the protocol in Protocol. For TCP, UDP and SCTP a Port field appears: pick a preset service, any port or Manual port…. For a port range, enter the first port and fill in Range end.
Add
Before you save, the form shows what the rule will do: New rule, Port will be added to the existing rule or No change — this port is already open, how many devices the source and destination cover, and a warning if the rule would open no traffic today. Click Add. The policy is compiled and applied right away; use Show on map in the notification to see the rule on the access map.
Preset services
| Service | Port |
|---|---|
| SSH | 22 |
| DNS | 53 |
| HTTP | 80 |
| HTTPS | 443 |
| SMB | 445 |
| MySQL | 3306 |
| RDP | 3389 |
| PostgreSQL | 5432 |
| Redis | 6379 |
| Kubernetes API | 6443 |
| MongoDB | 27017 |
Protocols
TCP, UDP, ICMP, SCTP, GRE, ESP, AH, IGMP. Ports can only be chosen for TCP, UDP and SCTP; for the others, the rule allows all traffic of that protocol.
Add ports to the same rule
There is one rule per source and destination. When you add a rule with the same source and destination again, the new port is added to the existing rule and shows up as a separate tag on it.
- To remove a port, click the × next to its tag.
- When the last port is removed, the rule is deleted too.
Delete a rule
Click Delete rule next to the rule and confirm. The access path closes immediately; this can't be undone.
How the policy is applied
On every change, the rules are compiled into a policy that's applied to the network's control server and reaches devices shortly after.
- When a change is applied, the panel shows the new policy version and the policy badge in the network header updates.
- If a change doesn't alter the policy, the rule list shows Policy already up to date. Saving group membership doesn't show that notice: when the policy is unchanged, the people-groups section stays silent.
See what a rule affects
- Rules with a label as the destination show how many devices they affect; if no machine carries the label, the rule affects no devices.
- A rule whose source is a group in which no member has a personal computer is marked covers no devices; that rule allows no traffic. Only network admins see the marker, and only on the source side.
What do members see?
Rules are read-only for people with the member role. Members only see rules that involve a group they're in: rules whose source is their group (what they can reach) and rules whose destination is their group (who can reach their devices). They don't see the network's other rules.