Zabbix plugin script for extracting uWSGI metrics from a log file

uwsgi_stats

#!/usr/bin/env python
"""
Extract metrics from a uWSGI log file

This should be saved as /etc/zabbix/pluginx/uwsgi_stats and called from
/etc/zabbix/zabbix_agentd.conf.d/uwsgi.conf, which should have contents:

    UserParameter=uwsgi.stats[*],/etc/zabbix/plugins/uwsgi_stats $1 $2

To gather these metrics in Zabbix, create a new item which calls this plugin and 
passes the filepath and metric key.  Eg:

    uwsgi.stats[/var/log/client/project/logs/prod/uwsgi.log,req_count]

The available metric keys are:

 - req_count -> requests per minute
 - get_req_count -> GET requests per minute
 - post_req_count -> POST requests per minute
 - avg_req_time -> average request time in ms
 - avg_req_size -> average request size in bytes

Author: David Winterbottom
"""

import datetime
import subprocess
import re
import sys
import os


def extract_metric(filepath, metric):
    # Determine time pattern to grep for
    now = datetime.datetime.now()

    # Fiddle with date format as uWSGI does not zero pad the day
    last_minute = now - datetime.timedelta(minutes=1)
    day = last_minute.strftime("%d").lstrip("0")
    month = last_minute.strftime("%b")
    time = last_minute.strftime("%H:%M")
    pattern = "%s  %s %s" % (month, day, time)

    # Extract data from access log
    try:
        output = subprocess.Popen(
            ["grep", pattern, filepath], stdout=subprocess.PIPE).communicate()[0]
    except subprocess.CalledProcessError:
        # Happens when no matches
        lines = []
    else:
        lines = output.split("\n")

    data = []
    regex = re.compile(r'.* (?P<method>[A-Z]+) (?P<path>\S+) => generated (?P<bytes>\d+) bytes in (?P<msecs>\d+) msecs \(HTTP/1.\d (?P<status>\d+).*')
    for line in lines:
        match = regex.match(line)
        if match:
            data.append(match.groupdict())

    if not data:
        return

    # Calculate and print appropriate metric
    if metric == 'avg_req_time':
        times = [int(x['msecs']) for x in data]
        return float(sum(times)) / len(times)
    elif metric == 'avg_req_size':
        sizes = [int(x['bytes']) for x in data]
        return float(sum(sizes)) / len(sizes)
    elif metric == 'req_count':
        return len(data)
    elif metric == 'get_req_count':
	return len([1 for x in data if x['method'] == 'GET'])
    elif metric == 'post_req_count':
	return len([1 for x in data if x['method'] == 'POST'])
    elif metric == '2xx_req_count':
	return len([1 for x in data if x['status'].startswith('2')])
    elif metric == '3xx_req_count':
	return len([1 for x in data if x['status'].startswith('3')])
    elif metric == '4xx_req_count':
	return len([1 for x in data if x['status'].startswith('4')])
    elif metric == '5xx_req_count':
	return len([1 for x in data if x['status'].startswith('5')])


if __name__ == '__main__':
    filepath = sys.argv[1]
    metric_key = sys.argv[2]
    metric = extract_metric(filepath, metric_key)
    if metric is not None:
        print metric