feat(zerotier): add ZeroTier network governance module

Add comprehensive ZeroTier integration for managing network access:

- Portal networks: manager-created ZeroTier network bindings
- Device registration: user-owned ZeroTier node endpoints
- Approval workflows: request/approve/revoke network access
- Activation sessions: time-limited network authorization
- Kill switch: emergency access revocation
- Reconciliation job: sync portal state with ZeroTier controller

Includes ZeroTier client SDK supporting both Central and self-hosted
controller APIs, with full CRUD operations for networks and members.
This commit is contained in:
2026-03-20 21:50:20 +10:30
parent 49e724222f
commit 1789590167
27 changed files with 4862 additions and 4 deletions
+186
View File
@@ -0,0 +1,186 @@
"""Device management service."""
import logging
import re
from gatehouse_app.extensions import db
from gatehouse_app.models import Device
from gatehouse_app.models.user import User
from gatehouse_app.services.audit_service import AuditService
from gatehouse_app.exceptions import (
DeviceNotFoundError,
DeviceAlreadyExistsError,
InvalidNodeIdError,
ValidationError,
)
logger = logging.getLogger(__name__)
_NODE_ID_RE = re.compile(r"^[0-9a-fA-F]{10}$")
def _validate_node_id(node_id: str) -> str:
node_id = node_id.strip().lower()
if not _NODE_ID_RE.match(node_id):
raise InvalidNodeIdError(
f"Invalid ZeroTier node ID '{node_id}' — must be exactly 10 hex characters."
)
return node_id
def register_device(
user_id: str,
organization_id: str,
node_id: str,
nickname: str | None = None,
hostname: str | None = None,
asset_tag: str | None = None,
serial_number: str | None = None,
) -> Device:
"""Register a new device for a user in an organization.
Args:
user_id: ID of the owning user
organization_id: ID of the organization
node_id: 10-char hex ZeroTier node ID
nickname: User-assigned friendly name
hostname: Device hostname
asset_tag: Corporate asset tag
serial_number: Device serial number
Returns:
The created Device record
"""
node_id = _validate_node_id(node_id)
existing = Device.query.filter(
Device.node_id == node_id,
Device.deleted_at.is_(None),
).first()
if existing:
raise DeviceAlreadyExistsError(
f"A device with node ID {node_id} is already registered."
)
device = Device(
user_id=user_id,
organization_id=organization_id,
node_id=node_id,
device_nickname=nickname,
hostname=hostname,
asset_tag=asset_tag,
serial_number=serial_number,
)
device.save()
AuditService.log_action(
action="device.registered",
user_id=user_id,
organization_id=organization_id,
resource_type="device",
resource_id=device.id,
metadata={"node_id": node_id, "nickname": nickname},
description=f"Device {node_id} registered",
success=True,
)
return device
def list_user_devices(user_id: str, organization_id: str) -> list[Device]:
"""List all active devices for a user in an organization."""
return Device.query.filter(
Device.user_id == user_id,
Device.organization_id == organization_id,
Device.deleted_at.is_(None),
).all()
def get_device(device_id: str, organization_id: str | None = None) -> Device:
"""Fetch a device by ID, optionally scoped to an organization."""
q = Device.query.filter(
Device.id == device_id,
Device.deleted_at.is_(None),
)
if organization_id:
q = q.filter(Device.organization_id == organization_id)
device = q.first()
if not device:
raise DeviceNotFoundError(f"Device {device_id} not found.")
return device
def get_device_by_node_id(node_id: str, organization_id: str | None = None) -> Device | None:
"""Find a device by its ZeroTier node ID, optionally scoped to org."""
node_id = _validate_node_id(node_id)
q = Device.query.filter(
Device.node_id == node_id,
Device.deleted_at.is_(None),
)
if organization_id:
q = q.filter(Device.organization_id == organization_id)
return q.first()
def update_device(
device_id: str,
user_id: str,
**kwargs,
) -> Device:
"""Update device fields. Only allows nickname, hostname, asset_tag."""
device = get_device(device_id)
if device.user_id != user_id:
raise DeviceNotFoundError("Device not found.")
allowed = {"device_nickname", "hostname", "asset_tag", "serial_number"}
for key in kwargs:
if key not in allowed:
raise ValidationError(f"Cannot update field: {key}")
device.update(**kwargs)
AuditService.log_action(
action="device.updated",
user_id=user_id,
organization_id=device.organization_id,
resource_type="device",
resource_id=device.id,
metadata=kwargs,
description=f"Device {device.node_id} updated",
success=True,
)
return device
def remove_device(device_id: str, user_id: str) -> None:
"""Soft-delete a device, deactivate its active memberships, and soft-delete all memberships.
Since memberships are device-centric (tied to a specific device/node_id), removing a device
means all its memberships are orphaned and must be cleaned up.
"""
device = get_device(device_id)
if device.user_id != user_id:
raise DeviceNotFoundError("Device not found.")
# Soft-delete all memberships (deactivates active ones first)
for membership in device.memberships:
if membership.deleted_at is None:
from gatehouse_app.services.network_access_service import revoke_membership_soft
revoke_membership_soft(membership.id, revoked_by_user_id=user_id)
device.delete(soft=True)
AuditService.log_action(
action="device.removed",
user_id=user_id,
organization_id=device.organization_id,
resource_type="device",
resource_id=device.id,
metadata={"node_id": device.node_id, "memberships_removed": len([m for m in device.memberships if m.deleted_at is None])},
description=f"Device {device.node_id} removed",
success=True,
)
File diff suppressed because it is too large Load Diff
@@ -0,0 +1,237 @@
"""Portal network management service."""
import logging
import re
from gatehouse_app.extensions import db
from gatehouse_app.models import PortalNetwork
from gatehouse_app.models.organization import Organization
from gatehouse_app.models.user import User
from gatehouse_app.services.audit_service import AuditService
from gatehouse_app.services import zerotier_api_service as zt
from gatehouse_app.utils.constants import NetworkRequestMode
from gatehouse_app.exceptions import (
NetworkNotFoundError,
InvalidNetworkIdError,
ValidationError,
ZeroTierNotFoundError,
ZeroTierAPIError,
)
logger = logging.getLogger(__name__)
_NETWORK_ID_RE = re.compile(r"^[0-9a-fA-F]{16}$")
def _validate_network_id(network_id: str) -> str:
network_id = network_id.strip().lower()
if not _NETWORK_ID_RE.match(network_id):
raise InvalidNetworkIdError(
f"Invalid ZeroTier network ID '{network_id}'"
"must be exactly 16 hex characters."
)
return network_id
def create_network(
organization_id: str,
name: str,
owner_user_id: str,
zerotier_network_id: str,
description: str | None = None,
environment: str | None = None,
request_mode: str = "approval_required",
default_activation_lifetime_minutes: int = 480,
max_activation_lifetime_minutes: int | None = None,
) -> PortalNetwork:
"""Create a new portal network.
Args:
organization_id: Owning organization
name: Human-readable name
owner_user_id: Manager who owns this network
zerotier_network_id: 16-char hex ZT network ID
description: Optional description
environment: production | staging | development | lab
request_mode: open | approval_required | invite_only
default_activation_lifetime_minutes: Default session length
max_activation_lifetime_minutes: Cap on activation lifetime
"""
from gatehouse_app.utils.constants import NetworkEnvironment
zerotier_network_id = _validate_network_id(zerotier_network_id)
existing = PortalNetwork.query.filter(
PortalNetwork.organization_id == organization_id,
PortalNetwork.zerotier_network_id == zerotier_network_id,
PortalNetwork.deleted_at.is_(None),
).first()
if existing:
raise ValidationError(
f"A portal network already exists for ZT network {zerotier_network_id} "
f"in this organization."
)
env = NetworkEnvironment(environment) if environment else NetworkEnvironment.DEVELOPMENT
mode = NetworkRequestMode(request_mode)
network = PortalNetwork(
organization_id=organization_id,
name=name,
description=description,
owner_user_id=owner_user_id,
zerotier_network_id=zerotier_network_id,
environment=env,
request_mode=mode,
default_activation_lifetime_minutes=default_activation_lifetime_minutes,
max_activation_lifetime_minutes=max_activation_lifetime_minutes,
)
network.save()
# Try to verify the network exists in ZeroTier
try:
zt_network = zt.get_network(zerotier_network_id)
logger.info(
f"[PortalNetwork] Verified ZT network {zerotier_network_id} "
f"exists in ZeroTier: {zt_network.name}"
)
except ZeroTierNotFoundError:
logger.warning(
f"[PortalNetwork] ZT network {zerotier_network_id} not found "
"in ZeroTier — will be reconciled later."
)
except ZeroTierAPIError as exc:
logger.warning(
f"[PortalNetwork] Could not verify ZT network {zerotier_network_id}: {exc}"
)
AuditService.log_action(
action="zt.network.created",
user_id=owner_user_id,
organization_id=organization_id,
resource_type="portal_network",
resource_id=network.id,
metadata={
"zerotier_network_id": zerotier_network_id,
"name": name,
"environment": env.value,
"request_mode": mode.value,
},
description=f"Portal network '{name}' created (ZT: {zerotier_network_id})",
success=True,
)
return network
def list_networks(
organization_id: str,
include_inactive: bool = False,
) -> list[PortalNetwork]:
"""List portal networks for an organization."""
q = PortalNetwork.query.filter(
PortalNetwork.organization_id == organization_id,
PortalNetwork.deleted_at.is_(None),
)
if not include_inactive:
q = q.filter(PortalNetwork.is_active.is_(True))
return q.all()
def get_network(network_id: str, organization_id: str | None = None) -> PortalNetwork:
"""Fetch a portal network by ID."""
q = PortalNetwork.query.filter(
PortalNetwork.id == network_id,
PortalNetwork.deleted_at.is_(None),
)
if organization_id:
q = q.filter(PortalNetwork.organization_id == organization_id)
network = q.first()
if not network:
raise NetworkNotFoundError(f"Portal network {network_id} not found.")
return network
def update_network(
network_id: str,
user_id: str,
**kwargs,
) -> PortalNetwork:
"""Update network metadata. Allowed: name, description, environment,
request_mode, default_activation_lifetime_minutes, max_activation_lifetime_minutes, is_active."""
network = get_network(network_id)
allowed = {
"name",
"description",
"environment",
"request_mode",
"default_activation_lifetime_minutes",
"max_activation_lifetime_minutes",
"is_active",
}
for key in kwargs:
if key not in allowed:
raise ValidationError(f"Cannot update field: {key}")
network.update(**kwargs)
AuditService.log_action(
action="zt.network.updated",
user_id=user_id,
organization_id=network.organization_id,
resource_type="portal_network",
resource_id=network.id,
metadata=kwargs,
description=f"Portal network '{network.name}' updated",
success=True,
)
return network
def delete_network(network_id: str, user_id: str) -> None:
"""Soft-delete a portal network and deactivate all memberships."""
network = get_network(network_id)
# Deauthorize all active memberships in ZeroTier
for membership in network.memberships:
if membership.deleted_at is None and membership.state.value == "active_authorized":
from gatehouse_app.services.network_access_service import deactivate_membership
deactivate_membership(membership.id, reason="network_deleted")
network.delete(soft=True)
AuditService.log_action(
action="zt.network.deleted",
user_id=user_id,
organization_id=network.organization_id,
resource_type="portal_network",
resource_id=network.id,
metadata={"zerotier_network_id": network.zerotier_network_id, "name": network.name},
description=f"Portal network '{network.name}' deleted",
success=True,
)
def get_network_members(network_id: str) -> list:
"""Return all DeviceNetworkMemberships for a network with user and device info."""
from gatehouse_app.models import DeviceNetworkMembership
return DeviceNetworkMembership.query.filter(
DeviceNetworkMembership.portal_network_id == network_id,
DeviceNetworkMembership.deleted_at.is_(None),
).all()
def get_network_pending_requests(network_id: str) -> list:
"""Return pending UserNetworkApprovals for a network."""
from gatehouse_app.models import UserNetworkApproval
from gatehouse_app.utils.constants import ApprovalState
return UserNetworkApproval.query.filter(
UserNetworkApproval.portal_network_id == network_id,
UserNetworkApproval.state == ApprovalState.PENDING,
UserNetworkApproval.deleted_at.is_(None),
).all()
@@ -0,0 +1,116 @@
"""ZeroTier API service — thin Flask adapter around the ZeroTierClient SDK.
Reads configuration from app config and translates SDK exceptions to
Secuird typed exceptions.
"""
import logging
from typing import Optional
from gatehouse_app.exceptions import ZeroTierAPIError
from gatehouse_app.utils.zerotier_client import (
APIMode,
ZeroTierAPIError as SDKZeroTierAPIError,
ZeroTierAuthError,
ZeroTierClient,
ZeroTierNotFoundError,
)
logger = logging.getLogger(__name__)
def _get_client(app=None) -> ZeroTierClient:
"""Build a ZeroTierClient from current app config."""
from flask import current_app
app = app or current_app
mode_str = app.config.get("ZEROTIER_API_MODE", "controller")
mode = APIMode.CENTRAL if mode_str == "central" else APIMode.CONTROLLER
return ZeroTierClient(
api_token=app.config.get("ZEROTIER_API_TOKEN", ""),
base_url=app.config.get("ZEROTIER_API_URL", "http://localhost:9993"),
mode=mode,
)
def get_status() -> dict:
"""Verify connectivity to the ZeroTier controller."""
client = _get_client()
try:
return client.get_status()
except SDKZeroTierAPIError as exc:
raise ZeroTierAPIError(str(exc), status_code=exc.status_code) from exc
def list_networks():
"""List all networks accessible to the configured token."""
client = _get_client()
try:
return client.list_networks()
except SDKZeroTierAPIError as exc:
raise ZeroTierAPIError(str(exc), status_code=exc.status_code) from exc
def get_network(network_id: str):
"""Fetch a single network by ID."""
client = _get_client()
try:
return client.get_network(network_id)
except SDKZeroTierAPIError as exc:
raise ZeroTierAPIError(str(exc), status_code=exc.status_code) from exc
def list_members(network_id: str):
"""List all members on a network."""
client = _get_client()
try:
return client.list_members(network_id)
except SDKZeroTierAPIError as exc:
raise ZeroTierAPIError(str(exc), status_code=exc.status_code) from exc
def get_member(network_id: str, node_id: str):
"""Fetch a single member on a network."""
client = _get_client()
try:
return client.get_member(network_id, node_id)
except SDKZeroTierAPIError as exc:
raise ZeroTierAPIError(str(exc), status_code=exc.status_code) from exc
def authorize_member(network_id: str, node_id: str):
"""Authorize a member on a network. Returns updated member."""
client = _get_client()
try:
return client.authorize_member(network_id, node_id)
except SDKZeroTierAPIError as exc:
raise ZeroTierAPIError(str(exc), status_code=exc.status_code) from exc
def deauthorize_member(network_id: str, node_id: str):
"""De-authorize a member on a network. Returns updated member."""
client = _get_client()
try:
return client.deauthorize_member(network_id, node_id)
except SDKZeroTierAPIError as exc:
raise ZeroTierAPIError(str(exc), status_code=exc.status_code) from exc
def add_member(network_id: str, node_id: str, authorized: bool = False):
"""Manually add/pre-provision a member on a network."""
client = _get_client()
try:
return client.add_member(network_id, node_id, authorized=authorized)
except SDKZeroTierAPIError as exc:
raise ZeroTierAPIError(str(exc), status_code=exc.status_code) from exc
def delete_network_member(network_id: str, node_id: str):
"""Remove a member entirely from a ZeroTier network."""
client = _get_client()
try:
return client.delete_member(network_id, node_id)
except SDKZeroTierAPIError as exc:
raise ZeroTierAPIError(str(exc), status_code=exc.status_code) from exc
@@ -0,0 +1,303 @@
"""ZeroTier reconciliation service — polling loop to sync state with the controller."""
import logging
from datetime import datetime, timezone
from gatehouse_app.extensions import db
from gatehouse_app.models import (
Device,
DeviceNetworkMembership,
ActivationSession,
ZeroTierMembership,
PortalNetwork,
UserNetworkApproval,
)
from gatehouse_app.services import zerotier_api_service as zt
from gatehouse_app.utils.constants import (
ActivationEndReason,
MembershipState,
ApprovalState,
)
logger = logging.getLogger(__name__)
def reconcile_expired_activations() -> int:
"""Find expired activation sessions and deactivate their memberships.
Returns the number of sessions expired.
"""
now = datetime.now(timezone.utc)
expired = ActivationSession.query.filter(
ActivationSession.expires_at < now,
ActivationSession.ended_at.is_(None),
ActivationSession.deleted_at.is_(None),
).all()
count = 0
for session in expired:
try:
_expire_session(session)
count += 1
except Exception as exc:
logger.error(f"[Reconciliation] Failed to expire session {session.id}: {exc}")
if count > 0:
logger.info(f"[Reconciliation] Expired {count} activation sessions.")
return count
def reconcile_network(portal_network_id: str) -> dict:
"""Full reconciliation for one portal network.
Returns a dict with counts of actions taken.
"""
network = PortalNetwork.query.get(portal_network_id)
if not network or not network.is_active:
return {"skipped": True, "reason": "network_inactive_or_deleted"}
zerotier_network_id = network.zerotier_network_id
actions = {
"zt_members_checked": 0,
"zt_members_added": 0,
"authorized": 0,
"deauthorized": 0,
"join_seen_updated": 0,
"unknown_members": [],
}
# Get current ZT members
try:
zt_members = {m.node_id: m for m in zt.list_members(zerotier_network_id)}
except Exception as exc:
logger.error(f"[Reconciliation] Failed to list ZT members for {zerotier_network_id}: {exc}")
actions["error"] = str(exc)
return actions
actions["zt_members_checked"] = len(zt_members)
# Get our portal memberships for this network
our_memberships = {
m.device.node_id: m
for m in DeviceNetworkMembership.query.filter(
DeviceNetworkMembership.portal_network_id == portal_network_id,
DeviceNetworkMembership.deleted_at.is_(None),
).all()
if m.device and m.device.deleted_at is None
}
# Reconcile each portal membership
for node_id, membership in our_memberships.items():
zt_member = zt_members.pop(node_id, None)
device = membership.device
if not zt_member:
# Member not seen in ZT yet
continue
actions["join_seen_updated"] += 1
# Update observed ZT membership
_sync_zt_membership(membership, zt_member)
# Sync authorization state
if membership.state == MembershipState.ACTIVE_AUTHORIZED:
if not zt_member.is_authorized:
# We think it's active but ZT says it's not — re-authorize
try:
zt.authorize_member(zerotier_network_id, node_id)
actions["authorized"] += 1
except Exception as exc:
logger.warning(f"[Reconciliation] Re-authorize failed for {node_id}: {exc}")
else:
if zt_member.is_authorized:
# We think it's not authorized but ZT says it is — deauthorize
# (could be manual override in ZT console)
try:
zt.deauthorize_member(zerotier_network_id, node_id)
actions["deauthorized"] += 1
except Exception as exc:
logger.warning(f"[Reconciliation] Deauthorize failed for {node_id}: {exc}")
# Unknown ZT members not in our portal
actions["unknown_members"] = list(zt_members.keys())
logger.info(
f"[Reconciliation] Network {zerotier_network_id}: "
f"checked={actions['zt_members_checked']} "
f"authorized={actions['authorized']} "
f"deauthorized={actions['deauthorized']} "
f"unknown={len(actions['unknown_members'])}"
)
return actions
def reconcile_all() -> dict:
"""Run reconciliation on all active portal networks.
Returns a summary dict.
"""
networks = PortalNetwork.query.filter(
PortalNetwork.is_active.is_(True),
PortalNetwork.deleted_at.is_(None),
).all()
results = {"networks_processed": 0, "errors": 0}
for network in networks:
try:
result = reconcile_network(network.id)
if "error" in result:
results["errors"] += 1
else:
results["networks_processed"] += 1
except Exception as exc:
logger.error(f"[Reconciliation] Failed to reconcile network {network.id}: {exc}")
results["errors"] += 1
deleted_result = reconcile_deleted_memberships()
results["deleted_memberships"] = deleted_result.get("deleted", 0)
results["delete_errors"] = deleted_result.get("errors", 0)
logger.info(
f"[Reconciliation] Complete: {results['networks_processed']} networks processed, "
f"{results['errors']} errors, {results.get('deleted_memberships', 0)} memberships purged."
)
return results
def reconcile_deleted_memberships() -> dict:
"""Find soft-deleted memberships and hard-delete them after ZeroTier cleanup.
Only processes memberships whose ZeroTier members are already de-authorized
(the de-authorize step happened in revoke_membership_soft). This function
removes the member from ZeroTier entirely and then hard-deletes the DB record.
"""
deleted = DeviceNetworkMembership.query.filter(
DeviceNetworkMembership.deleted_at.isnot(None),
).all()
if not deleted:
return {"deleted": 0, "errors": 0}
results = {"deleted": 0, "errors": 0}
for membership in deleted:
try:
device = Device.query.get(membership.device_id)
network = PortalNetwork.query.get(membership.portal_network_id)
if not device or not network:
db.session.delete(membership)
db.session.commit()
results["deleted"] += 1
continue
try:
zt.delete_network_member(network.zerotier_network_id, device.node_id)
logger.info(f"[Reconciliation] Deleted {device.node_id} from ZT network {network.zerotier_network_id}")
except Exception as zt_exc:
logger.warning(
f"[Reconciliation] ZT delete failed for {device.node_id} "
f"on {network.zerotier_network_id}: {zt_exc}"
)
db.session.delete(membership)
db.session.commit()
results["deleted"] += 1
except Exception as exc:
logger.error(f"[Reconciliation] Failed to hard-delete membership {membership.id}: {exc}")
results["errors"] += 1
if results["deleted"] > 0:
logger.info(f"[Reconciliation] Purged {results['deleted']} memberships.")
return results
def _sync_zt_membership(membership: DeviceNetworkMembership, zt_member) -> None:
"""Update the ZeroTierMembership cache record from a ZT API response."""
device = membership.device
network = membership.portal_network
zt_membership = ZeroTierMembership.query.filter(
ZeroTierMembership.zerotier_network_id == network.zerotier_network_id,
ZeroTierMembership.node_id == device.node_id,
ZeroTierMembership.deleted_at.is_(None),
).first()
if not zt_membership:
zt_membership = ZeroTierMembership(
organization_id=membership.organization_id,
device_network_membership_id=membership.id,
zerotier_network_id=network.zerotier_network_id,
node_id=device.node_id,
)
zt_membership.member_seen = True
zt_membership.authorized = zt_member.is_authorized
zt_membership.last_synced_at = datetime.now(timezone.utc)
zt_membership.raw_controller_payload = zt_member.raw
if zt_member.last_seen and zt_member.last_seen > 0:
zt_membership.join_seen_at = datetime.fromtimestamp(
zt_member.last_seen / 1000, tz=timezone.utc
)
zt_membership.save()
# Update membership join_seen flag
if not membership.join_seen:
membership.join_seen = True
membership.state = MembershipState.JOINED_DEAUTHORIZED
membership.save()
def _expire_session(session: ActivationSession) -> None:
"""Expire an activation session and deauthorize the membership in ZT."""
session.ended_at = datetime.now(timezone.utc)
session.end_reason = ActivationEndReason.EXPIRED
session.save()
membership = DeviceNetworkMembership.query.get(session.device_network_membership_id)
if membership:
membership.state = MembershipState.ACTIVATION_EXPIRED
membership.currently_authorized = False
membership.save()
device = Device.query.get(membership.device_id)
network = PortalNetwork.query.get(membership.portal_network_id)
if device and network:
try:
zt.deauthorize_member(network.zerotier_network_id, device.node_id)
# Update ZT membership cache
zt_membership = ZeroTierMembership.query.filter(
ZeroTierMembership.zerotier_network_id == network.zerotier_network_id,
ZeroTierMembership.node_id == device.node_id,
ZeroTierMembership.deleted_at.is_(None),
).first()
if zt_membership:
zt_membership.authorized = False
zt_membership.last_synced_at = datetime.now(timezone.utc)
zt_membership.save()
except Exception as exc:
logger.warning(
f"[_expire_session] Failed to deauthorize {device.node_id} "
f"on {network.zerotier_network_id}: {exc}"
)
from gatehouse_app.services.audit_service import AuditService
AuditService.log_action(
action="zt.activation.expired",
user_id=session.user_id,
organization_id=session.organization_id,
resource_type="activation_session",
resource_id=session.id,
metadata={"membership_id": session.device_network_membership_id},
description="Activation session expired",
success=True,
)