website_watchdog

#!/usr/bin/env python

import os, sys, httplib, logging
from urlparse import urlparse
import boto.ec2

# Configure this to a rotated log file
LOG_FILE = '/var/log/httpd/website_watchdog.log'

# Configure your keys here
AWS_ACCESS_KEY = 'XXX'
AWS_SECRET_KEY = 'YYY'

# Configure the name of the custom cloudwatch metric
CW_METRIC_NAME = 'webserver-response'

#  All the servers should be listed here with the elements:
#     Server ID, Server Name, URL, Snippet
aSERVERS = [
    ( 'i-1111111', 'Production', 'https://www.yourdomain.com/page', 'some sample text in page'),
    ]

oBotoEc2 = boto.connect_cloudwatch(aws_access_key_id=AWS_ACCESS_KEY, aws_secret_access_key=AWS_SECRET_KEY)

def postCloudWatch (sServerId, sServerDesc, sStatus):
    sNamespace = 'EC2: ' + sServerDesc
    logging.info('Posting to EC2 CloudWatch: namespace: %s, metric: %s, instance: %s, value: %s' %
                 (sNamespace, CW_METRIC_NAME, sServerId, sStatus))
    # put_metric_data(namespace, name, value=None, timestamp=None, unit=None, dimensions=None, statistics=None)
    oBotoEc2.put_metric_data(sNamespace, CW_METRIC_NAME, value=sStatus,
                             timestamp=None, unit='Count', dimensions= { 'InstanceId' : [ sServerId ] } )
                                 

def getSiteContent (sUrl):
    try: 
        oUrl = urlparse(sUrl)
        if (oUrl.scheme == 'https'):
            oConnect = httplib.HTTPConnection(oUrl.netloc)
        else:
            oConnect = httplib.HTTPSConnection(oUrl.netloc)
        oConnect.request('GET', oUrl.path)
        oResponse = oConnect.getresponse()
        if (oResponse.status in [ 200, 302 ] ):
            mReturn = oResponse.read()
        else:
            logging.error('URL did not return valid response: %s - status: %d' % (sUrl, oResponse.status))
            mReturn = False
        oConnect.close()
        return mReturn
    except:
        logging.error('Failed getting response from URL: ' + sUrl)
        oConnect.close()
        return False

def validateSite (sId, sServerDesc, sUrl, sSnippet):
    logging.info('Validating site: %s - %s - %s - %s' % (sId, sServerDesc, sUrl, sSnippet))
    sContent = getSiteContent(sUrl)
    if (sContent != False and sSnippet in sContent):
        # logging.info('Site appears valid - match found in: ' + sUrl)
        postCloudWatch(sId, sServerDesc, '1')
    else:
        if (sContent != False):
            logging.error('No match found in URL: %s  Match:%s' % (sUrl, sSnippet))
        postCloudWatch(sId, sServerDesc, '0')
    
def main():
    logging.basicConfig(level=logging.INFO, filename=LOG_FILE, 
            format='%(asctime)s %(levelname)s: %(message)s', 
            datefmt='%Y-%m-%d %H:%M:%S')

    for sId, sServerDesc, sUrl, sSnippet in aSERVERS:
        validateSite(sId, sServerDesc, sUrl, sSnippet)

if __name__ == '__main__':
    main()