argusclient -- A minimal client library for Argus webservice

This is a minimal and thin layer of Python client code on top of the Argus webservices REST API. Most of the library API is 1:1 with that of REST API so it serves to be more of a convenience than an abstraction. This means you still need to be familiar with the underlying REST API to be effective. For more information on the REST API and data model, refer to the Argus - User Guide. Special thanks to Demian Brecht for giving a lot of feedback early and helping to shape the API and the project.

You can also browse the Python API documentation online at: https://salesforce.github.io/python-argusclient/

A quick primer to using argusclient

Below you will find functional and self-explanatory code that shows how to do the following:

Import the relevant pieces from argusclient
Create the main entry point and establish login session
Query for existing namespaces
Create a new namespace
Collect metrics and annotations
Post metrics and annotations
Query for existing dashboards
Update or Create dashboard
Query for existing alerts
Delete alert
Create an alert along with a trigger and a notification

In addition, also look at the bundled example named splunk_to_argus.py that shows how to extract metrics from Splunk and push them to Argus.

Some package imports and initializations that we use later

import sys, os, time, calendar, getpass, logging, random
import lxml.etree

from argusclient import *
from argusclient.dashboardtags import DASHBOARD, CHART, TITLE, METRIC, FLAGS

logging.basicConfig()
logging.root.setLevel(logging.INFO)
logging.getLogger("requests").setLevel(logging.WARN)

endpoint = "http://localhost:8080/argusws"
user = "hdara"
password = None

tags = { "host": "hdara-wsl" }
fields = { "user": user }
curtm = long(calendar.timegm(time.gmtime()))*1000
ns_name = "hdara-ns"
ns_access_addl_users = ("hdara",)
dashboard_name = "hdara.test.dashboard"
alert_name = "hdara.test.alert"
scope_name = "hdara"
metric_name = "test"
ans = []

argus = ArgusServiceClient(user,
                           password or getpass.getpass("SSO password for %s: " % user),
                           endpoint=endpoint)
logging.info("Logging in")
argus.login()

Check if a namespace exists and create one if missing

logging.info("Looking up existing namespace with name: %s", ns_name)
nss = dict((ns.qualifier, ns) for ns in argus.namespaces.values())
ns = nss.get(ns_name)
if not ns:
    logging.info("Creating new namespace with name: %s", ns_name)
    ns = argus.namespaces.add(Namespace(ns_name))

Generate some random metrics against hdara-ns:hdara:test and mark the start and end with annotations.

logging.info("Generating some metric and annotation data for the dashboard")
m = Metric(scope_name, metric_name, tags=tags, namespace=ns_name)
for t in xrange(10, 0, -1):
    # Warden requires 1 minute gap between successive data points.
    ts = curtm-t*60*1000
    m.datapoints[ts] = random.randint(50, 100)
    if not ans or t == 1:
        ans.append(Annotation("script", "hdara", "test", ts, ts, "generated", tags=tags, fields=dict(event=ans and "start" or "end", **fields)))

Send metrics and annotations to Argus

logging.info("Adding metrics data to Argus")
am_resp = argus.metrics.add([m]);
if am_resp.error_count():
    logging.error("Errors reported in metric data: errorCount: %s errorMessages: %s", am_resp.error_count(), am_resp.error_messages())
logging.info("Adding annotation data to Argus")
an_resp = argus.annotations.add(ans)
if an_resp.error_count():
    logging.error("Errors reported in annotation data: errorCount: %s errorMessages: %s", an_resp.error_count(), an_resp.error_messages())

Generate dashboard content

mquery = str(MetricQuery(scope_name, metric_name, "sum", tags=tags, stTimeSpec="-1d", enTimeSpec="-0d", namespace=ns_name))
aquery = str(AnnotationQuery(scope_name, metric_name, "generated", tags=tags, stTimeSpec="-1d", enTimeSpec="-0d"))
content = lxml.etree.tostring(DASHBOARD(
    CHART(
        TITLE("hdara.test"),
        METRIC(mquery, name="hdara.test.metric"),
        FLAGS(aquery, name="hdara.test.annotation"),
        name="Chart"
        )
), method="html")
dashbobj.content = content

Update or Create dashboard

logging.info("Looking up existing dashboard with name: %s", dashboard_name)
dashbobj = argus.dashboards.get_user_dashboard(user, dashboard_name, shared=False)
if not dashbobj:
    logging.info("Creating new dashboard with name: %s", dashboard_name)
    dashbobj = Dashboard(dashboard_name, content, shared=True, description="A new dashboard")
    dashbobj = argus.dashboards.add(dashbobj)
else:
    logging.info("Updating dashboard with name: %s id %s", dashboard_name, dashbobj.argus_id)
    dashbobj.content = content
    argus.dashboards.update(dashbobj.argus_id, dashbobj)
logging.info("Dashboard url: %s", os.path.join(os.path.dirname(endpoint), "argus/#/dashboards", str(dashbobj.argus_id)).replace("-ws", "-ui"))

Look for an existing alert and delete it so that we can recreate it

logging.info("Looking up existing alert with name: %s owned by user: %s", alert_name, user)
alertobj = argus.alerts.get_user_alert(user, alert_name, shared=False)
if alertobj:
    logging.info("Deleting existing alert with name: %s id: %s", alert_name, alertobj.argus_id)
    argus.alerts.delete(alertobj.argus_id)

Finally, create alert with a trigger and a notification

logging.info("Creating new alert with alert name: %s", alert_name)
alertobj = argus.alerts.add(Alert(alert_name, mquery, "* */1 * * *",
                                  trigger=Trigger("hdara.test.trigger", Trigger.GREATER_THAN, 100000, 600000),
                                  notification=Notification("hdara.test.notification", Notification.EMAIL, subscriptions=["[email protected]"]),
                                  shared=True))

HTTP not found on Alert Obj

I faced this in the alert object but I suspect it occurs across all objects.

Issue: Per documentation, the check to see if alert exists only works if the alert exists.
If it doesn't, it throws httplib.NOT_FOUND(404) exception from here. Traceback below.

All actions including Alert creation/modification should be idempotent. So DELETE and PUT is the way to go. We just need a better mechanism for handling the error.
May be instead of raise, just catch the exception and gently let the calling function know? Thoughts?

Exception:

Traceback (most recent call last):
  File "alert.py", line 32, in <module>
    alertobj = argus.alerts.get_user_alert(user, alert_name, shared=False)
  File "client.py", line 473, in get_user_alert
    alerts = self.argus._request("get", "alerts/meta", params=dict(alertname=alertName, ownername=ownerName, shared=shared))
  File "client.py", line 555, in with_retry
    return f(*args, **kwargs)
  File "client.py", line 588, in with_auth_token
    return f(*args, **kwargs)
  File "client.py", line 708, in _request
    return self._request_no_auth(method, path, params, dataObj, encCls, decCls)
  File "client.py", line 730, in _request_no_auth
    res = check_success(resp, decCls)
  File "client.py", line 744, in check_success
    raise ArgusException("Object not found at endpoint: %s message: %s" % (resp.request.url, resp.text))
argusclient.client.ArgusException: Object not found at endpoint: 
https://<argus_url>/argusws/alerts/meta?shared=False&ownername=<owner_name>&alertname=<alert_name> 
message: {"status":404,"message":"Not Found"}

salesforce / python-argusclient Goto Github PK

python-argusclient's Introduction

argusclient -- A minimal client library for Argus webservice

A quick primer to using argusclient

Some package imports and initializations that we use later

Login to the service and establish session

Check if a namespace exists and create one if missing

Generate some random metrics against hdara-ns:hdara:test and mark the start and end with annotations.

Send metrics and annotations to Argus

Generate dashboard content

Update or Create dashboard

Look for an existing alert and delete it so that we can recreate it

Finally, create alert with a trigger and a notification

python-argusclient's People

Contributors

Stargazers

Watchers

Forkers

python-argusclient's Issues

Recommend Projects

Recommend Topics

Recommend Org