Simple nginx stats aggregation without NGINX Plus, Lua or NginScript - all you need is Python3!

parse_nginx_stats.py

#!/usr/bin/env python3

# Start this script before you start NGINX, so it is ready and waiting for log data.
# You should start it in the background with & or run it from supervisord or something.
# If the port is above 1024, you can run this script as a non-root user (recommended).
#
# It listens for raw Syslog data over UDP, parses is, aggregates it, and writes
# it to a file every flush_interval seconds.  The output file format looks like this:
#  key<TAB>value
# Here is an example (note that the delimiter is actually TAB, not a space):
'''
count 6
200 1
500 5
2xx 1
5xx 5
'''
# Add this to your nginx config in the section that you want to collect log data on (http, server, location):
#   log_format  status '$status\n';
#   access_log  syslog:server=127.0.0.1:10514,nohostname,tag=nginx status;
#

import sys
import time
import socket

if len(sys.argv) < 2:
    report_file ="/tmp/nginx_status"
else:
    report_file = sys.argv[1]

listen = ("127.0.0.1", 10514)
flush_interval = 5
total_msgs = 0
last_flush = 0
rx_buffer = 1024
delimiter = "\t"
statuses = {}
started_at = time.gmtime()

# Clear report file when we first start
with open(report_file, "w") as fh:
    print(delimiter.join( ("started", time.strftime("%a, %d %b %Y %H:%M:%S +0000", started_at)) ), file=fh)
    print(delimiter.join( ("updated", time.strftime("%a, %d %b %Y %H:%M:%S +0000")) ), file=fh)
    print(delimiter.join( ("count", "0") ), file=fh)

sock = socket.socket(socket.AF_INET, socket.SOCK_DGRAM)
sock.bind(listen)

# For each message, record stats
while True:
    message, addr = sock.recvfrom(rx_buffer)

    total_msgs += 1
    msg_len = len(message)

    if msg_len < 5:
        continue

    try:
        status = int(message[msg_len-4:msg_len-1])
    except ValueError:
        continue;

    if status in statuses:
        statuses[status] += 1
    else:
        statuses[status] = 1

    now = time.time()

    if now - last_flush > flush_interval:
        groupstatuses = {
            "2xx": 0,
            "3xx": 0,
            "4xx": 0,
            "5xx": 0,
        };

        with open(report_file, "w") as fh:
            print(delimiter.join( ("started", time.strftime("%a, %d %b %Y %H:%M:%S +0000", started_at)) ), file=fh)
            print(delimiter.join( ("updated", time.strftime("%a, %d %b %Y %H:%M:%S +0000")) ), file=fh)
            print(delimiter.join( ("count", str(total_msgs)) ), file=fh)

            for status in statuses:
                # Write exact status stats
                print(delimiter.join( (str(status), str(statuses[status])) ), file=fh)
                if status < 200 or status >= 600:
                    continue

                # Record group status (2xx, 3xx, etc)
                groupstatus = "%sxx" % str(status)[0]
                groupstatuses[groupstatus] += statuses[status]

            for groupstatus in groupstatuses:
                # Write group status stats

                print(delimiter.join( (groupstatus, str(groupstatuses[groupstatus])) ), file=fh)