✨ — No need cron tasks for agents watching (fix #76)

CHANGELOG.md

@@ -5,6 +5,7 @@
 - ✨ — Allow to customize agent User-Agent header (#78)
 - 📝 — Document how to add data to requests (#77)
 - ✨ — No need cron tasks for DB cleaning anymore (#74 and #75)
+- ✨ — No need cron tasks for agents watching (#76)
 
 ## 0.7.4
 

argos/commands.py

@@ -140,47 +140,6 @@ def start(host, port, config, reload):
     uvicorn.run("argos.server:app", host=host, port=port, reload=reload)
 
 
-def validate_time_without_agent(ctx, param, value):
-    if value <= 0:
-        raise click.BadParameter("Should be a positive integer")
-    return value
-
-
-@server.command()
-@click.option(
-    "--time-without-agent",
-    default=5,
-    help="Time without seeing an agent after which a warning will be issued, in minutes. "
-    "Default is 5 minutes.",
-    callback=validate_time_without_agent,
-)
-@click.option(
-    "--config",
-    default="argos-config.yaml",
-    help="Path of the configuration file. "
-    "If ARGOS_YAML_FILE environment variable is set, its value will be used instead.",
-    envvar="ARGOS_YAML_FILE",
-    callback=validate_config_access,
-)
-@coroutine
-async def watch_agents(time_without_agent, config):
-    """Watch agents (to run routinely)
-
-    Issues a warning if no agent has been seen by the server for a given time.
-    """
-    # It’s mandatory to do it before the imports
-    os.environ["ARGOS_YAML_FILE"] = config
-
-    # The imports are made here otherwise the agent will need server configuration files.
-    from argos.server import queries
-
-    db = await get_db()
-    agents = await queries.get_recent_agents_count(db, time_without_agent)
-    if agents == 0:
-        click.echo(f"No agent has been seen in the last {time_without_agent} minutes.")
-        sysexit(1)
-
-
 @server.command(short_help="Load or reload tasks’ configuration")
 @click.option(
     "--config",
@@ -537,6 +496,7 @@ async def test_mail(config, domain, severity):
             check="body-contains",
             expected="foo",
             frequency=1,
+            ip_version=4,
             selected_by="test",
             selected_at=now,
         )
@@ -601,6 +561,7 @@ async def test_gotify(config, domain, severity):
             check="body-contains",
             expected="foo",
             frequency=1,
+            ip_version=4,
             selected_by="test",
             selected_at=now,
         )
@@ -668,6 +629,7 @@ async def test_apprise(config, domain, severity, apprise_group):
             check="body-contains",
             expected="foo",
             frequency=1,
+            ip_version=4,
             selected_by="test",
             selected_at=now,
         )

argos/config-example.yaml

@@ -102,6 +102,10 @@ general:
       - local
     unknown:
       - local
+    # This alert is triggered when no Argos agent has been seen in a while
+    # See recurring_tasks.time_without_agent below
+    no_agent:
+      - local
   # Mail configuration is quite straight-forward
   # mail:
   #   mailfrom: no-reply@example.org
@@ -145,15 +149,19 @@ ssl:
     - "1d": critical
     - "5d": warning
 
-# Argos will do some cleaning in the background for you
+# Argos will execute some tasks in the background for you
 # every 2 minutes and needs some configuration for that
-cleaning:
+recurring_tasks:
   # Max number of results per tasks you want to keep
   # Minimum value is 1, default is 100
   max_results: 100
   # Max number of seconds a task can be locked
   # Minimum value is 61, default is 100
   max_lock_seconds: 100
+  # Max number of minutes without seing an agent
+  # before sending an alert
+  # Minimum value is 1, default is 5
+  time_without_agent: 5
 
 # It's also possible to define the checks in another file
 # with the include syntax:

argos/schemas/config.py

@@ -48,9 +48,10 @@ class SSL(BaseModel):
     thresholds: List[Annotated[Tuple[int, Severity], BeforeValidator(parse_threshold)]]
 
 
-class Cleaning(BaseModel):
+class RecurringTasks(BaseModel):
     max_results: int
     max_lock_seconds: int
+    time_without_agent: int
 
     @field_validator("max_results", mode="before")
     def parse_max_results(cls, value):
@@ -68,6 +69,14 @@ class Cleaning(BaseModel):
 
         return 100
 
+    @field_validator("time_without_agent", mode="before")
+    def parse_time_without_agent(cls, value):
+        """Ensure that time_without_agent is at least one minute"""
+        if value >= 1:
+            return value
+
+        return 5
+
 
 class WebsiteCheck(BaseModel):
     key: str
@@ -211,6 +220,7 @@ class Alert(BaseModel):
     warning: List[str]
     critical: List[str]
     unknown: List[str]
+    no_agent: List[str]
 
 
 class GotifyUrl(BaseModel):
@@ -285,5 +295,5 @@ class Config(BaseModel):
     general: General
     service: Service
     ssl: SSL
-    cleaning: Cleaning
+    recurring_tasks: RecurringTasks
     websites: List[Website]

argos/server/alerting.py

@@ -74,6 +74,91 @@ def get_icon_from_severity(severity: str) -> str:
     return icon
 
 
+def send_mail(mail: EmailMessage, config: Mail):
+    """Send message by mail"""
+
+    if config.ssl:
+        logger.debug("Mail notification: SSL")
+        context = ssl.create_default_context()
+        smtp = smtplib.SMTP_SSL(host=config.host, port=config.port, context=context)
+    else:
+        smtp = smtplib.SMTP(
+            host=config.host,  # type: ignore
+            port=config.port,
+        )
+        if config.starttls:
+            logger.debug("Mail notification: STARTTLS")
+            context = ssl.create_default_context()
+            smtp.starttls(context=context)
+
+    if config.auth is not None:
+        logger.debug("Mail notification: authentification")
+        smtp.login(config.auth.login, config.auth.password)
+
+    for address in config.addresses:
+        logger.debug("Sending mail to %s", address)
+        logger.debug(mail.get_body())
+        smtp.send_message(mail, to_addrs=address)
+
+
+def send_gotify_msg(config, payload):
+    """Send message with gotify"""
+    headers = {"accept": "application/json", "content-type": "application/json"}
+
+    for url in config:
+        logger.debug("Sending gotify message(s) to %s", url.url)
+        for token in url.tokens:
+            try:
+                res = httpx.post(
+                    f"{url.url}message",
+                    params={"token": token},
+                    headers=headers,
+                    json=payload,
+                )
+                res.raise_for_status()
+            except httpx.RequestError as err:
+                logger.error(
+                    "An error occurred while sending a message to %s with token %s",
+                    err.request.url,
+                    token,
+                )
+
+
+def no_agent_alert(config: Config):
+    """Alert"""
+    msg = "You should check what’s going on with your Argos agents."
+    twa = config.recurring_tasks.time_without_agent
+    if twa > 1:
+        subject = f"No agent has been seen within the last {twa} minutes"
+    else:
+        subject = "No agent has been seen within the last minute"
+
+    if "local" in config.general.alerts.no_agent:
+        logger.error(subject)
+
+    if config.general.mail is not None and "mail" in config.general.alerts.no_agent:
+        mail = EmailMessage()
+        mail["Subject"] = f"[Argos] {subject}"
+        mail["From"] = config.general.mail.mailfrom
+        mail.set_content(msg)
+        send_mail(mail, config.general.mail)
+
+    if config.general.gotify is not None and "gotify" in config.general.alerts.no_agent:
+        priority = 9
+        payload = {"title": subject, "message": msg, "priority": priority}
+        send_gotify_msg(config.general.gotify, payload)
+
+    if config.general.apprise is not None:
+        for notif_way in config.general.alerts.no_agent:
+            if notif_way.startswith("apprise:"):
+                group = notif_way[8:]
+                apobj = apprise.Apprise()
+                for channel in config.general.apprise[group]:
+                    apobj.add(channel)
+
+                apobj.notify(title=subject, body=msg)
+
+
 def handle_alert(config: Config, result, task, severity, old_severity, request):  # pylint: disable-msg=too-many-positional-arguments
     """Dispatch alert through configured alert channels"""
 
@@ -163,36 +248,13 @@ See results of task on {request.url_for('get_task_results_view', task_id=task.id
     ] = f"[Argos] {icon} {urlparse(task.url).netloc} (IPv{task.ip_version}): status {severity}"
     mail["From"] = config.mailfrom
     mail.set_content(msg)
-
+    send_mail(mail, config)
-    if config.ssl:
-        logger.debug("Mail notification: SSL")
-        context = ssl.create_default_context()
-        smtp = smtplib.SMTP_SSL(host=config.host, port=config.port, context=context)
-    else:
-        smtp = smtplib.SMTP(
-            host=config.host,  # type: ignore
-            port=config.port,
-        )
-        if config.starttls:
-            logger.debug("Mail notification: STARTTLS")
-            context = ssl.create_default_context()
-            smtp.starttls(context=context)
-
-    if config.auth is not None:
-        logger.debug("Mail notification: authentification")
-        smtp.login(config.auth.login, config.auth.password)
-
-    for address in config.addresses:
-        logger.debug("Sending mail to %s", address)
-        logger.debug(msg)
-        smtp.send_message(mail, to_addrs=address)
 
 
 def notify_with_gotify(  # pylint: disable-msg=too-many-positional-arguments
     result, task, severity: str, old_severity: str, config: List[GotifyUrl], request
 ) -> None:
     logger.debug("Will send gotify notification")
-    headers = {"accept": "application/json", "content-type": "application/json"}
 
     icon = get_icon_from_severity(severity)
     priority = 9
@@ -228,20 +290,4 @@ See results of task on <{request.url_for('get_task_results_view', task_id=task.i
 
     payload = {"title": subject, "message": msg, "priority": priority, "extras": extras}
 
-    for url in config:
+    send_gotify_msg(config, payload)
-        logger.debug("Sending gotify message(s) to %s", url.url)
-        for token in url.tokens:
-            try:
-                res = httpx.post(
-                    f"{url.url}message",
-                    params={"token": token},
-                    headers=headers,
-                    json=payload,
-                )
-                res.raise_for_status()
-            except httpx.RequestError as err:
-                logger.error(
-                    "An error occurred while sending a message to %s with token %s",
-                    err.request.url,
-                    token,
-                )

argos/server/main.py

@@ -13,6 +13,7 @@ from sqlalchemy.orm import sessionmaker
 
 from argos.logging import logger, set_log_level
 from argos.server import models, routes, queries
+from argos.server.alerting import no_agent_alert
 from argos.server.exceptions import NotAuthenticatedException, auth_exception_handler
 from argos.server.settings import read_yaml_config
 
@@ -128,20 +129,24 @@ def create_manager(cookie_secret: str) -> LoginManager:
 
 
 @repeat_every(seconds=120, logger=logger)
-async def cleanup() -> None:
+async def recurring_tasks() -> None:
+    """Recurring DB cleanup and watch-agents tasks"""
     set_log_level("info", quiet=True)
-    logger.info("Start DB cleanup tasks.")
+    logger.info("Start background recurring tasks")
     with app.state.SessionLocal() as db:
-        removed = await queries.remove_old_results(
+        config = app.state.config.recurring_tasks
-            db, app.state.config.cleaning.max_results
+        removed = await queries.remove_old_results(db, config.max_results)
-        )
-        updated = await queries.release_old_locks(
-            db, app.state.config.cleaning.max_lock_seconds
-        )
-
         logger.info("%i results removed", removed)
+
+        updated = await queries.release_old_locks(db, config.max_lock_seconds)
         logger.info("%i locks released", updated)
 
+        agents = await queries.get_recent_agents_count(db, config.time_without_agent)
+        if agents == 0:
+            no_agent_alert(app.state.config)
+
+    logger.info("Background recurring tasks ended")
+
 
 @asynccontextmanager
 async def lifespan(appli: FastAPI):
@@ -159,7 +164,7 @@ async def lifespan(appli: FastAPI):
             "There is no tasks in the database. "
             'Please launch the command "argos server reload-config"'
         )
-    await cleanup()
+    await recurring_tasks()
 
     yield
 

docs/cli.md

@@ -94,7 +94,6 @@ Commands:
   test-gotify      Send a test gotify notification
   test-mail        Send a test email
   user             User management
-  watch-agents     Watch agents (to run routinely)
 ```
 
 <!--[[[end]]]
@@ -151,33 +150,6 @@ Options:
 -->
 
 
-### Server watch-agents
-
-<!--
-.. [[[cog
-    help(["server", "watch-agents", "--help"])
-.. ]]] -->
-
-```man
-Usage: argos server watch-agents [OPTIONS]
-
-  Watch agents (to run routinely)
-
-  Issues a warning if no agent has been seen by the server for a given time.
-
-Options:
-  --time-without-agent INTEGER  Time without seeing an agent after which a
-                                warning will be issued, in minutes. Default is 5
-                                minutes.
-  --config TEXT                 Path of the configuration file. If
-                                ARGOS_YAML_FILE environment variable is set, its
-                                value will be used instead.
-  --help                        Show this message and exit.
-```
-
-<!--[[[end]]]
--->
-
 ### Server reload-config
 
 <!--

docs/developer/dependencies.md

@@ -14,7 +14,9 @@ description: Many thanks to their developers!
 - [Alembic](https://alembic.sqlalchemy.org) is used for DB migrations;
 - [Tenacity](https://github.com/jd/tenacity) a small utility to retry a function in case an error occured;
 - [Uvicorn](https://www.uvicorn.org/) is the tool used to run our server;
-- [Gunicorn](https://gunicorn.org/) is the recommended WSGI HTTP server for production.
+- [Gunicorn](https://gunicorn.org/) is the recommended WSGI HTTP server for production;
+- [Apprise](https://github.com/caronc/apprise/wiki) allows Argos to send notifications through a lot of channels;
+- [FastAPI Utilities](https://fastapiutils.github.io/fastapi-utils/) is in charge of recurring tasks.
 
 ## CSS framework
 

tests/config.yaml

@@ -26,6 +26,8 @@ general:
       - local
     unknown:
       - local
+    no_agent:
+      - local
 service:
   secrets:
     # Secrets can be generated using `argos server generate-token`.
@@ -38,15 +40,19 @@ ssl:
     - "1d": critical
     - "5d": warning
 
-# Argos will do some cleaning in the background for you
+# Argos will execute some tasks in the background for you
 # every 2 minutes and needs some configuration for that
-cleaning:
+recurring_tasks:
   # Max number of results per tasks you want to keep
   # Minimum value is 1, default is 100
   max_results: 100
   # Max number of seconds a task can be locked
   # Minimum value is 61, default is 100
   max_lock_seconds: 100
+  # Max number of seconds without seing an agent
+  # before sending an alert
+  # Minimum value is 61, default is 300
+  time_without_agent: 300
 
 # It's also possible to define the checks in another file
 # with the include syntax:

tests/test_queries.py

@@ -235,6 +235,7 @@ def empty_config():
                 warning=["", ""],
                 critical=["", ""],
                 unknown=["", ""],
+                no_agent=["", ""],
             ),
         ),
         service=schemas.config.Service(
@@ -243,9 +244,10 @@ def empty_config():
             ]
         ),
         ssl=schemas.config.SSL(thresholds=[]),
-        cleaning=schemas.config.Cleaning(
+        recurring_tasks=schemas.config.RecurringTasks(
             max_results=100,
             max_lock_seconds=120,
+            time_without_agent=300,
         ),
         websites=[],
     )