> For the complete documentation index, see [llms.txt](https://wiki.datadike.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://wiki.datadike.com/pam/deploy-operate/network-requirements.md).

# Network and connectivity requirements

The connectivity the solution itself needs — who reaches the console and on which ports, what the solution originates outbound, how to publish it safely, and why it needs no internet access of its own

## What this page covers

The firewall rules the **solution** needs: what must reach it, what it must reach, and in which direction.

It is not the list of ports used to reach *assets*. That list is per platform and lives in [Ports and protocols](https://wiki.datadike.com/pam/reference/ports.md) — write your asset-side firewall requests from there, and use this page for the solution's own rules.

{% hint style="danger" %}
Once deployed, the solution itself requires **no internet access at all**. Block it.

Deploy it in a **segmented, segregated network**. If your network is flat, the solution cannot deliver what it is for — regularise the network design before you deploy it, not after.
{% endhint %}

## Traffic the solution needs

| Source                 | Destination     | Service                        | Port    | Why                                                                                                                                                      |
| ---------------------- | --------------- | ------------------------------ | ------- | -------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Console users          | The solution    | HTTP                           | 80      | Console access over HTTP                                                                                                                                 |
| Console users          | The solution    | HTTPS                          | 443     | Console access over HTTPS                                                                                                                                |
| Console users          | The solution    | SSH                            | 2222    | Command-line access to the solution                                                                                                                      |
| The solution           | Web-access host | SSH                            | 22      | The solution connects to the host to deploy the agent and the control mechanism                                                                          |
| The solution           | Web-access host | RDP                            | 3389    | As above, where the host is reached over RDP                                                                                                             |
| Web-access host        | The solution    | HTTPS                          | 443     | The host authenticates back to the solution                                                                                                              |
| Web-access host        | The solution    | HTTP                           | 80      | As above, over HTTP                                                                                                                                      |
| Web-access host        | Internet        | Depends on the sites published | Depends | Only the URLs actually published in the solution — HTTP, HTTPS, or a non-standard port such as 8080. Enumerate the published URLs and open exactly those |
| Your monitoring server | The solution    | Zabbix agent                   | 10050   | Zabbix agent v1, where you monitor the node with Zabbix                                                                                                  |
| The solution           | Windows asset   | WinRM                          | 5985    | Remote execution of automations (account collection, credential rotation)                                                                                |

{% hint style="info" %}
Two notes on that table.

**HTTP and HTTPS are both listed.** Whether you actually need port 80 depends on whether the instance is meant to answer over HTTP at all. Confirm before you open it — an instance reached only over HTTPS does not need it.

**The WinRM rule belongs to the solution, not to your monitoring server.** The source table attributes it to the Zabbix server, which appears to be a copy-paste of the row above it: automations are executed by the solution's own task service. See [Ports used by automations](https://wiki.datadike.com/pam/reference/ports.md#ports-used-by-automations).
{% endhint %}

## The web-access host is the exception to "no internet"

Two different machines, and confusing them produces either a broken feature or an unnecessary hole in the perimeter:

* **The solution** needs no internet access. Block it outbound.
* **The web-access host** — the host that runs published web sessions — needs outbound access to the sites you publish, and nothing else. Open it per URL.

## Publishing the console outside the perimeter

If users must reach the console from outside, publish it through **NAT behind a WAF**, over **port 443 (HTTPS)** only. From that single entry point the user reaches every internal asset — RDP, SSH, Telnet, VNC, databases — with no direct connectivity between the user's device and the protected assets.

{% hint style="warning" %}
Do not publish the console to the internet without a firewall and a WAF in front of it. If that protection is not in place, allow access only over a controlled VPN instead.
{% endhint %}

## Networks the solution cannot route to

Two mechanisms, in opposite directions. An asset uses one or the other and never both — the asset record forces the choice.

|                                                                               | Direction                | What the network must allow                                                                                                                                                     |
| ----------------------------------------------------------------------------- | ------------------------ | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| [Proxy](https://wiki.datadike.com/pam/administration/connectivity/proxy.md)   | Solution → asset network | A route from the solution to the proxy host, with the SSH port open in that direction. The proxy needs a stored credential, and must itself be able to see the assets it serves |
| [Broker](https://wiki.datadike.com/pam/administration/connectivity/broker.md) | Asset network → broker   | Outbound **443/tcp** from the remote environment to the broker. No inbound port is opened in the remote network at all                                                          |

{% hint style="danger" %}
A broker must be reached by a **DNS name, never a bare IP address**. The address is written into the certificates and into the enrolment tokens, and cannot be changed afterwards without reinstalling the broker. An IP literal only ever works for one network path: behind NAT, the LAN and the internet do not see the same address.
{% endhint %}

Gateways and connectors communicate in **one direction only**, chosen per environment: allow outbound traffic or allow inbound traffic, not both. RDP and SSH traffic can be encapsulated in HTTPS through the secure tunnel. See [Gateways](https://wiki.datadike.com/pam/administration/connectivity/gateways.md).

## Traffic the solution originates

| Destination                   | What for                                                  | Port                                                                                                                                                                                                                                                              |
| ----------------------------- | --------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Assets                        | User sessions and automations                             | Per platform and per asset — see [Ports and protocols](https://wiki.datadike.com/pam/reference/ports.md)                                                                                                                                                          |
| Your syslog or SIEM collector | Audit event forwarding                                    | Set in the appliance configuration file `/opt/datadike/config/config.txt`, as `SYSLOG_ADDR=<host>:<port>` with `SYSLOG_FACILITY`. The documented example uses port 514 and facility `local2`. Changing it requires a support ticket and a restart of the services |
| Your mail server              | Backup delivery, alerts and approval workflows            | Not documented — take it from your mail server's configuration                                                                                                                                                                                                    |
| Your directory                | LDAP / Active Directory authentication and account import | Not documented — take it from your directory. For LDAPS, the certificate files are loaded onto the appliance and the service restarted. See [LDAP and Active Directory](https://wiki.datadike.com/pam/administration/identities/ldap-and-ad.md)                   |

## Monitoring

Point your monitor at the unauthenticated health endpoint `/api/health/`. It returns JSON with `status`, `db_status`, `db_time`, `redis_status`, `redis_time` and `time`, so a successful poll proves the database and Redis are reachable — not merely that the web server is answering. In a cluster, poll the virtual address and each node directly.

## Clusters

A cluster is fronted by a **single virtual address**, and clients are pointed at that address rather than at any individual node. Each node also answers directly on its own address, which is what makes per-node monitoring possible — so your monitoring source needs to reach each node, not only the virtual address.

The ports used **between** cluster nodes for replication come from DataDike. Get them before you place the nodes in different network segments.

## Common errors

| Symptom                                                                                             | Cause                                                                                                  | Fix                                                                                                                                                                                        |
| --------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------ | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| The console does not answer on its hostname during initial setup                                    | The DNS record does not exist yet                                                                      | Add a temporary entry to the hosts file of your own workstation to get through setup, then create the record in your private DNS — and in public DNS too if you are publishing through NAT |
| The firewall rule is open but the session still fails                                               | The rule was written from the platform's default port, while the asset uses a different effective port | Check the port on the asset record. See [Ports and protocols](https://wiki.datadike.com/pam/reference/ports.md)                                                                            |
| Reverse access through a broker stops working after a network change, or never worked from one side | The broker was enrolled against an IP literal                                                          | The broker address must be a DNS name. Changing it means reinstalling the broker                                                                                                           |
| SSH to the solution on 2222 is refused                                                              | The port is closed, or the containers are not running on the node                                      | Test reachability on port 2222 from the client network, then check the state of the containers on the node                                                                                 |
| Every one-time password is rejected                                                                 | The node's clock has drifted; TOTP is validated to the second                                          | Synchronise the clock over NTP and set the correct timezone                                                                                                                                |

## Related pages

* [System requirements](https://wiki.datadike.com/pam/deploy-operate/requirements.md)
* [Ports and protocols](https://wiki.datadike.com/pam/reference/ports.md)
* [Broker](https://wiki.datadike.com/pam/administration/connectivity/broker.md)
* [Gateways](https://wiki.datadike.com/pam/administration/connectivity/gateways.md)
* [Proxy](https://wiki.datadike.com/pam/administration/connectivity/proxy.md)
