added notes about cronjob for ha

# Conflicts:
#	README.md

.DEBIAN/control

@@ -2,7 +2,7 @@ Package: fastapi-dls
 Version: 0.0
 Architecture: all
 Maintainer: Oscar Krause oscar.krause@collinwebdesigns.de
-Depends: python3, python3-fastapi, python3-uvicorn, python3-dotenv, python3-dateutil, python3-jose, python3-sqlalchemy, python3-pycryptodome, python3-markdown, uvicorn, openssl
+Depends: python3, python3-fastapi, python3-uvicorn, python3-dotenv, python3-dateutil, python3-jose, python3-sqlalchemy, python3-pycryptodome, python3-markdown, python3-httpx, uvicorn, openssl
 Recommends: curl
 Installed-Size: 10240
 Homepage: https://git.collinwebdesigns.de/oscar.krause/fastapi-dls

README.md

@@ -32,6 +32,17 @@ Tested with Ubuntu 22.10 (from Proxmox templates), actually its consuming 100mb
 
 - Make sure your timezone is set correct on you fastapi-dls server and your client
 
+**HA Setup Notes**
+
+- only *failover mode* is supported by team-green (see *high availability* in official user guide)
+- make sure you're using same configuration on each node
+- use same `instance.private.pem` and `instance.private.key` on each node
+- add `cronjob` on each node with `curl -X GET --insecure https://localhost/-/ha/replicate`
+
+If you want to use *real* HA, you should use a proxy in front of this service and use a clustered database in backend.
+This is not documented and supported by me, but it *can* work. Please ask the community for help.
+Maybe the simplest solution for HA-ing this service is to use a Docker-Swarm with redundant storage and database.
+
 ## Docker
 
 Docker-Images are available here:
@@ -371,22 +382,27 @@ After first success you have to replace `--issue` with `--renew`.
 # Configuration
 
 | Variable               | Default                                | Usage                                                                                                              |
-|------------------------|----------------------------------------|------------------------------------------------------------------------------------------------------|
+|------------------------|----------------------------------------|--------------------------------------------------------------------------------------------------------------------|
 | `DEBUG`                | `false`                                | Toggles `fastapi` debug mode                                                                                       |
 | `DLS_URL`              | `localhost`                            | Used in client-token to tell guest driver where dls instance is reachable                                          |
 | `DLS_PORT`             | `443`                                  | Used in client-token to tell guest driver where dls instance is reachable                                          |
+| `HA_REPLICATE`         |                                        | `DLS_URL` + `DLS_PORT` of primary DLS instance, e.g. `dls-node:443` (for HA only **two** nodes are supported!) \*1 |
+| `HA_ROLE`              |                                        | `PRIMARY` or `SECONDARY`                                                                                           |
 | `TOKEN_EXPIRE_DAYS`    | `1`                                    | Client auth-token validity (used for authenticate client against api, **not `.tok` file!**)                        |
 | `LEASE_EXPIRE_DAYS`    | `90`                                   | Lease time in days                                                                                                 |
-| `LEASE_RENEWAL_PERIOD` | `0.15`                                 | The percentage of the lease period that must elapse before a licensed client can renew a license \*1 |
+| `LEASE_RENEWAL_PERIOD` | `0.15`                                 | The percentage of the lease period that must elapse before a licensed client can renew a license \*2               |
 | `DATABASE`             | `sqlite:///db.sqlite`                  | See [official SQLAlchemy docs](https://docs.sqlalchemy.org/en/14/core/engines.html)                                |
-| `CORS_ORIGINS`         | `https://{DLS_URL}`                    | Sets `Access-Control-Allow-Origin` header (comma separated string) \*2                               |
+| `CORS_ORIGINS`         | `https://{DLS_URL}`                    | Sets `Access-Control-Allow-Origin` header (comma separated string) \*3                                             |
 | `SITE_KEY_XID`         | `00000000-0000-0000-0000-000000000000` | Site identification uuid                                                                                           |
 | `INSTANCE_REF`         | `10000000-0000-0000-0000-000000000001` | Instance identification uuid                                                                                       |
 | `ALLOTMENT_REF`        | `20000000-0000-0000-0000-000000000001` | Allotment identification uuid                                                                                      |
-| `INSTANCE_KEY_RSA`     | `<app-dir>/cert/instance.private.pem`  | Site-wide private RSA key for singing JWTs \*3                                                       |
-| `INSTANCE_KEY_PUB`     | `<app-dir>/cert/instance.public.pem`   | Site-wide public key \*3                                                                             |
+| `INSTANCE_KEY_RSA`     | `<app-dir>/cert/instance.private.pem`  | Site-wide private RSA key for singing JWTs \*4                                                                     |
+| `INSTANCE_KEY_PUB`     | `<app-dir>/cert/instance.public.pem`   | Site-wide public key \*4                                                                                           |
 
-\*1 For example, if the lease period is one day and the renewal period is 20%, the client attempts to renew its license
+\*1 If you want to use HA, this value should be point to `secondary` on `primary` and `primary` on `secondary`. Don't
+use same database for both instances!
+
+\*2 For example, if the lease period is one day and the renewal period is 20%, the client attempts to renew its license
 every 4.8 hours. If network connectivity is lost, the loss of connectivity is detected during license renewal and the
 client has 19.2 hours in which to re-establish connectivity before its license expires.
 

app/main.py

@@ -6,7 +6,7 @@ from os.path import join, dirname
 from os import getenv as env
 
 from dotenv import load_dotenv
-from fastapi import FastAPI
+from fastapi import FastAPI, BackgroundTasks
 from fastapi.requests import Request
 from json import loads as json_loads
 from datetime import datetime, timedelta
@@ -19,7 +19,7 @@ from starlette.responses import StreamingResponse, JSONResponse as JSONr, HTMLRe
 from sqlalchemy import create_engine
 from sqlalchemy.orm import sessionmaker
 
-from util import load_key, load_file
+from util import load_key, load_file, ha_replicate
 from orm import Origin, Lease, init as db_init, migrate
 
 load_dotenv('../version.env')
@@ -36,6 +36,7 @@ db_init(db), migrate(db)
 # everything prefixed with "INSTANCE_*" is used as "SERVICE_INSTANCE_*" or "SI_*" in official dls service
 DLS_URL = str(env('DLS_URL', 'localhost'))
 DLS_PORT = int(env('DLS_PORT', '443'))
+HA_REPLICATE, HA_ROLE = env('HA_REPLICATE', None), env('HA_ROLE', None)  # only failover is supported
 SITE_KEY_XID = str(env('SITE_KEY_XID', '00000000-0000-0000-0000-000000000000'))
 INSTANCE_REF = str(env('INSTANCE_REF', '10000000-0000-0000-0000-000000000001'))
 ALLOTMENT_REF = str(env('ALLOTMENT_REF', '20000000-0000-0000-0000-000000000001'))
@@ -199,6 +200,36 @@ async def _client_token():
     cur_time = datetime.utcnow()
     exp_time = cur_time + CLIENT_TOKEN_EXPIRE_DELTA
 
+    if HA_REPLICATE is not None and HA_ROLE.lower() == "secondary":
+        return RedirectResponse(f'https://{HA_REPLICATE}/-/client-token')
+
+    idx_port, idx_node = 0, 0
+
+    def create_svc_port_set(port: int):
+        idx = idx_port
+        return {
+            "idx": idx,
+            "d_name": "DLS",
+            "svc_port_map": [{"service": "auth", "port": port}, {"service": "lease", "port": port}]
+        }
+
+    def create_node_url(url: str, svc_port_set_idx: int):
+        idx = idx_node
+        return {"idx": idx, "url": url, "url_qr": url, "svc_port_set_idx": svc_port_set_idx}
+
+    service_instance_configuration = {
+        "nls_service_instance_ref": INSTANCE_REF,
+        "svc_port_set_list": [create_svc_port_set(DLS_PORT)],
+        "node_url_list": [create_node_url(DLS_URL, idx_port)]
+    }
+    idx_port += 1
+    idx_node += 1
+
+    if HA_REPLICATE is not None and HA_ROLE.lower() == "primary":
+        SEC_URL, SEC_PORT, *invalid = HA_REPLICATE.split(':')
+        service_instance_configuration['svc_port_set_list'].append(create_svc_port_set(SEC_PORT))
+        service_instance_configuration['node_url_list'].append(create_node_url(SEC_URL, idx_port))
+
     payload = {
         "jti": str(uuid4()),
         "iss": "NLS Service Instance",
@@ -209,17 +240,7 @@ async def _client_token():
         "update_mode": "ABSOLUTE",
         "scope_ref_list": [ALLOTMENT_REF],
         "fulfillment_class_ref_list": [],
-        "service_instance_configuration": {
-            "nls_service_instance_ref": INSTANCE_REF,
-            "svc_port_set_list": [
-                {
-                    "idx": 0,
-                    "d_name": "DLS",
-                    "svc_port_map": [{"service": "auth", "port": DLS_PORT}, {"service": "lease", "port": DLS_PORT}]
-                }
-            ],
-            "node_url_list": [{"idx": 0, "url": DLS_URL, "url_qr": DLS_URL, "svc_port_set_idx": 0}]
-        },
+        "service_instance_configuration": service_instance_configuration,
         "service_instance_public_key_configuration": {
             "service_instance_public_key_me": {
                 "mod": hex(INSTANCE_KEY_PUB.public_key().n)[2:],
@@ -239,6 +260,67 @@ async def _client_token():
     return response
 
 
+@app.get('/-/ha/replicate', summary='* HA replicate - trigger')
+async def _ha_replicate_to_ha(request: Request, background_tasks: BackgroundTasks):
+    if HA_REPLICATE is None or HA_ROLE is None:
+        logger.warning('HA replicate endpoint triggerd, but no value for "HA_REPLICATE" or "HA_ROLE" is set!')
+        return JSONr(status_code=503, content={'status': 503, 'detail': 'no value for "HA_REPLICATE" or "HA_ROLE" set'})
+
+    session = sessionmaker(bind=db)()
+    origins = [origin.serialize() for origin in session.query(Origin).all()]
+    leases = [lease.serialize(renewal_period=LEASE_RENEWAL_PERIOD, renewal_delta=LEASE_RENEWAL_DELTA) for lease in session.query(Lease).all()]
+
+    background_tasks.add_task(ha_replicate, logger, HA_REPLICATE, HA_ROLE, VERSION, DLS_URL, DLS_PORT, SITE_KEY_XID, INSTANCE_REF, origins, leases)
+    return JSONr(status_code=202, content=None)
+
+
+@app.put('/-/ha/replicate', summary='* HA replicate')
+async def _ha_replicate_by_ha(request: Request):
+    j, cur_time = json_loads((await request.body()).decode('utf-8')), datetime.utcnow()
+
+    if HA_REPLICATE is None:
+        logger.warning(f'HA replicate endpoint triggerd, but no value for "HA_REPLICATE" is set!')
+        return JSONr(status_code=503, content={'status': 503, 'detail': 'no value for "HA_REPLICATE" set'})
+
+    version = j.get('VERSION')
+    if version != VERSION:
+        logger.error(f'Version missmatch on HA replication task!')
+        return JSONr(status_code=503, content={'status': 503, 'detail': 'Missmatch for "VERSION"'})
+
+    site_key_xid = j.get('SITE_KEY_XID')
+    if site_key_xid != SITE_KEY_XID:
+        logger.error(f'Site-Key missmatch on HA replication task!')
+        return JSONr(status_code=503, content={'status': 503, 'detail': 'Missmatch for "SITE_KEY_XID"'})
+
+    instance_ref = j.get('INSTANCE_REF')
+    if instance_ref != INSTANCE_REF:
+        logger.error(f'Version missmatch on HA replication task!')
+        return JSONr(status_code=503, content={'status': 503, 'detail': 'Missmatch for "INSTANCE_REF"'})
+
+    sync_timestamp, max_seconds_behind = datetime.fromisoformat(j.get('sync_timestamp')), 30
+    if sync_timestamp <= cur_time - timedelta(seconds=max_seconds_behind):
+        logger.error(f'Request time more than {max_seconds_behind}s behind!')
+        return JSONr(status_code=503, content={'status': 503, 'detail': 'Request time behind'})
+
+    origins, leases = j.get('origins'), j.get('leases')
+    for origin in origins:
+        origin_ref = origin.get('origin_ref')
+        logging.info(f'> [    ha    ]: origin {origin_ref}')
+        data = Origin.deserialize(origin)
+        Origin.create_or_update(db, data)
+
+    for lease in leases:
+        lease_ref = lease.get('lease_ref')
+        x = Lease.find_by_lease_ref(db, lease_ref)
+        if x is not None and x.lease_updated > sync_timestamp:
+            continue
+        logging.info(f'> [    ha    ]: lease {lease_ref}')
+        data = Lease.deserialize(lease)
+        Lease.create_or_update(db, data)
+
+    return JSONr(status_code=202, content=None)
+
+
 # venv/lib/python3.9/site-packages/nls_services_auth/test/test_origins_controller.py
 @app.post('/auth/v1/origin', description='find or create an origin')
 async def auth_v1_origin(request: Request):
@@ -545,6 +627,22 @@ async def app_on_startup():
     Your client-token file (.tok) is valid for {str(CLIENT_TOKEN_EXPIRE_DELTA)}.
     ''')
 
+    if HA_REPLICATE is not None and HA_ROLE is not None:
+        from hashlib import sha1
+
+        sha1digest = sha1(INSTANCE_KEY_RSA.export_key()).hexdigest()
+        fingerprint_key = ':'.join(sha1digest[i: i + 2] for i in range(0, len(sha1digest), 2))
+        sha1digest = sha1(INSTANCE_KEY_PUB.export_key()).hexdigest()
+        fingerprint_pub = ':'.join(sha1digest[i: i + 2] for i in range(0, len(sha1digest), 2))
+
+        logger.info(f'''
+    HA mode is enabled. Make sure theses fingerprints matches on all your nodes:
+      - INSTANCE_KEY_RSA: "{str(fingerprint_key)}"
+      - INSTANCE_KEY_PUB: "{str(fingerprint_pub)}"
+    
+    This node ({HA_ROLE}) listens to "https://{DLS_URL}:{DLS_PORT}" and replicates to "https://{HA_REPLICATE}".
+        ''')
+
 
 if __name__ == '__main__':
     import uvicorn

app/orm.py

@@ -32,6 +32,16 @@ class Origin(Base):
             'os_version': self.os_version,
         }
 
+    @staticmethod
+    def deserialize(j) -> "Origin":
+        return Origin(
+            origin_ref=j.get('origin_ref'),
+            hostname=j.get('hostname'),
+            guest_driver_version=j.get('guest_driver_version'),
+            os_platform=j.get('os_platform'),
+            os_version=j.get('os_version'),
+        )
+
     @staticmethod
     def create_statement(engine: Engine):
         from sqlalchemy.schema import CreateTable
@@ -95,6 +105,16 @@ class Lease(Base):
             'lease_renewal': lease_renewal.isoformat(),
         }
 
+    @staticmethod
+    def deserialize(j) -> "Lease":
+        return Lease(
+            lease_ref=j.get('lease_ref'),
+            origin_ref=j.get('origin_ref'),
+            lease_created=datetime.fromisoformat(j.get('lease_created')),
+            lease_expires=datetime.fromisoformat(j.get('lease_expires')),
+            lease_updated=datetime.fromisoformat(j.get('lease_updated')),
+        )
+
     @staticmethod
     def create_statement(engine: Engine):
         from sqlalchemy.schema import CreateTable

app/util.py

@@ -26,3 +26,29 @@ def generate_key() -> "RsaKey":
         from Cryptodome.PublicKey.RSA import RsaKey
 
     return RSA.generate(bits=2048)
+
+
+def ha_replicate(logger: "logging.Logger", ha_replicate: str, ha_role: str, version: str, dls_url: str, dls_port: int, site_key_xid: str, instance_ref: str, origins: list, leases: list) -> bool:
+    from datetime import datetime
+    import httpx
+
+    if f'{dls_url}:{dls_port}' == ha_replicate:
+        logger.error(f'Failed to replicate this node ({ha_role}) to "{ha_replicate}": can\'t replicate to itself')
+        return False
+
+    data = {
+        'VERSION': str(version),
+        'HA_REPLICATE': f'{dls_url}:{dls_port}',
+        'SITE_KEY_XID': str(site_key_xid),
+        'INSTANCE_REF': str(instance_ref),
+        'origins': origins,
+        'leases': leases,
+        'sync_timestamp': datetime.utcnow().isoformat(),
+    }
+
+    r = httpx.put(f'https://{ha_replicate}/-/ha/replicate', json=data, verify=False)
+    if r.status_code == 202:
+        logger.info(f'Successfully replicated this node ({ha_role}) to "{ha_replicate}".')
+        return True
+    logger.error(f'Failed to replicate this node ({ha_role}) to "{ha_replicate}": {r.status_code} - {r.content}')
+    return False

requirements.txt

@@ -6,3 +6,4 @@ python-dateutil==2.8.2
 sqlalchemy==2.0.3
 markdown==3.4.1
 python-dotenv==0.21.1
+httpx==0.23.3