build-node-stats

#!/usr/bin/env python
from commands import getstatusoutput
from socket import gethostname
from argparse import ArgumentParser
import time
import pickle
import struct
import socket

def chunks(l, n):
  for i in xrange(0, len(l), n):
    yield l[i:i+n]

def format(s, **kwds):
  return s % kwds

CARBON_SERVER="localhost"
CARBON_PORT=2004

# 100 metrics at the time
def sendMetrics(metrics):
  for l in chunks(metrics, 100):
    payload = pickle.dumps(l)
    print len(payload)
    header = struct.pack("!L", len(payload))
    message = header + payload
    sock = socket.socket()
    print CARBON_SERVER, CARBON_PORT
    sock.connect((CARBON_SERVER, CARBON_PORT))
    sock.sendall(message)
    sock.close()

def aggregate(metrics):
  aggregated = {}
  for x in metrics:
    print x[0]
    if x[0] in aggregated.keys():
      aggregated[x[0]][1][1] += x[1][1]
      print x[0],"now", aggregated[x[0]][1][1]
    else:
      aggregated[x[0]] = x
  return [x for x in aggregated.values()]

if __name__ == "__main__":
  parser = ArgumentParser()
  parser.add_argument("--dest", dest="server", help="destination carbon server", default="localhost:2004")
  opts = parser.parse_args() 
  if not ":" in opts.server:
    opts.dest += ":2004"
  
  CARBON_SERVER, CARBON_PORT = opts.server.split(":", 1)
  CARBON_PORT=int(CARBON_PORT)
  
  fields = ("pid", "comm", "pcpu", "rss", "vsize")
  cmd = 'ps -U cmsbuild -o %s= -o %s= -o %s= -o %s= -o %s=' % fields
  err, out = getstatusoutput(cmd)
  if err:
    print "Error while reading process data"
    exit(1)
  print out
  metrics = []
  timestamp = time.time()
  for l in out.split("\n"):
    data = dict(zip(fields, [x for x in l.split(" ") if x]))
    try:
      data["pid"] = int(data["pid"])
      data["pcpu"] = float(data["pcpu"])
      data["rss"] = float(data["rss"])
      data["vsize"] = float(data["vsize"])
    except:
      continue
    if data["pcpu"] < 1:
      continue
    hostname = gethostname().split(".",1)[0]
    timestamp = time.time()
    metrics.append([".".join(["RunStats", hostname, data["comm"], "pcpu"]), [timestamp, data["pcpu"]]])
    metrics.append([".".join(["RunStats", hostname, data["comm"], "rss"]), [timestamp, data["rss"]]])
    metrics.append([".".join(["RunStats", hostname, data["comm"], "vsize"]), [timestamp, data["vsize"]]])
  sendMetrics(aggregate(metrics))