Python /PostgresqSQL 9.4 – Server Performance Data Capture V.2

In this second version, the scripts which capture data for the servers are using classes. In addition the the script that captures dynamic data from the remote servers captures additional data. In specific it captures RAM, HD usage in raw numbers and not only “humanized” format. The humanized fields were not taken out. As a result the appropriate tables had to be modified, which means the schema has changed as well. Also, the database and the scripts are now being installed via puppet. The Puppet manifests are far from polished. They need more work but they do work.

These scripts gather static and dynamic information from servers and insert that data into a PostgresSQL database. The static information is information that unless a major upgrade takes place it never changes. They Dynamic data is performance data of the servers. The purpose of the static data is to be able to query for dynamic information which being inserted to the database every X amount of minutes via cron.

These scripts work only on Linux/Unix based machines.

The PUPPET modules used to install the Database and the scripts are located here

The static information for the remote servers are as follows :

hostname
iface
ipaddress
OS
OSRel
OSKern
total_M
brand
Hz
cores
arch

The dynamic information for the remote servers are as follows :

hostname
iface
ipaddress
total_ram_hum
used_ram_hum
total_ram_raw
used_ram_raw
used_ram_perc
total_HD_hum
used_HD_hum
total_HD_raw
used_HD_raw
used_HD_perc
cpu_use_perc
swap_used_hum
swap_total_hum
swap_perc
swap_used_raw
swap_total_raw

The static.py script will need to be run only one time on the remote servers or when a major upgrade occurs that might change configuration regarding RAM, Partitions, IP, Operating System (even an upgrade), CPU, NIC replacement.

The dynamic.py script will be run on the remote servers via cron. It is the script that captures information which is being constantly changed, like memory, storage, swap usage. All this data is sent for insertion to the remote database. The script executes it self via cron and then sent to be inserted into a PostgreSQL database.

In both scripts data is entered into a dictionary and then a connection to the database is created in order to insert the data.

The Static.py script

#!/usr/bin/python
import psutil
import os
import math
import platform
import subprocess
import socket
import psycopg2
import netifaces as ni
import humanize
from cpuinfo import cpuinfo


class Static():
    def __init__(self):
        #NIC INFO
        self.hostname   = socket.gethostname()
        self.iface      = ni.interfaces()[1]
        self.ipaddress  = ni.ifaddresses(self.iface)[2][0]['addr']
        #---OS INFO
        #For Linux (RH-Debian) Operating Systems
        self.distro  = platform.dist()[0]
        self.release = platform.dist()[1]
        self.kernel  = platform.release()
        #For Windows Operating Systems
        self.osinfo_2_os    = platform.uname()[0]
        self.osinfo_2_ver   = platform.uname()[2]
        self.osinfo_2_rel   = platform.uname()[3]
        #----RAM INFO
        raw_totalM = psutil.virtual_memory().total
        self.total_M    = humanize.naturalsize(raw_totalM)
        #----CPU INFO
        self.info       = cpuinfo.get_cpu_info()
        self.brand      = self.info['brand']
        self.Hz         = self.info['hz_advertised']
        self.cores        = self.info['count']
        self.arch       = self.info['bits']

    def get_OS_make(self):
       if platform.system()  =="Linux":
           return self.distro, self.release, self.kernel
       elif platform.system()     =="Windows":
           return self.osinfo_2_os, self.osinfo_2_ver, self.osinfo_2_rel

info = Static()



hostname  = info.hostname
iface     = info.iface
ipaddress = info.ipaddress
OS        = info.get_OS_make()[0]
OSRel     = info.get_OS_make()[1]
OSKern    = info.get_OS_make()[2]
total_M   = info.total_M
brand     = info.brand
Hz        = info.Hz
cores     = info.cores
arch      = info.arch




#Create the Database PostgreSQL 9.4 connection.
conn = psycopg2.connect("host='172.31.98.161' dbname='servers' user='seeker'")
cur = conn.cursor() #Create the cursor
#Create a Dictionary to pass the value of each function.
server_info = {'hostname': hostname, 'iface':iface, 'ipaddress': ipaddress, 'OS': OS, 'OSRel': OSRel, 'OSKern': OSKern, 'total_M': total_M, 'brand': brand, 'Hz':Hz, 'cores': cores, 'arch': arch}
cur.execute("INSERT INTO servers(hostname, iface, ipaddress, OS, OSRel, OSKern, total_M, brand, Hz, cores, arch) VALUES ('%s','%s','%s','%s','%s','%s','%s','%s','%s','%s','%s')" % (hostname, iface, ipaddress, OS, OSRel, OSKern, total_M, brand, Hz, cores, arch))
#If this is not present the changes will not get commited.
conn.commit()

The Dynamic.py srcipt

#THIS FILE IS MANAGED BY PUPPET
#!/usr/bin/python
import psutil
import os
import math
import platform
import subprocess
import socket
import psycopg2
import netifaces as ni
import humanize
from cpuinfo import cpuinfo

class Dynamic():
    def __init__(self):
        #NIC INFO
        self.hostname   = socket.gethostname()
        self.iface      = ni.interfaces()[1]
        self.ipaddress  = ni.ifaddresses(self.iface)[2][0]['addr']
        #RAM USAGE INFO-------------------------------------------------------------------
        self.total_ram_hum       = humanize.naturalsize(psutil.virtual_memory().total)
        self.used_ram_hum        = humanize.naturalsize(psutil.virtual_memory().used)
        #---------Raw info
        self.total_ram_raw       = (psutil.virtual_memory().total)
        self.used_ram_raw        = (psutil.virtual_memory().used)
        self.used_ram_perc       = psutil.virtual_memory().percent
        #HD USAGE INFO-------------------------------------------------------------------
        self.total_HD_hum        = humanize.naturalsize(psutil.disk_usage('/').total)
        self.used_HD_hum         = humanize.naturalsize(psutil.disk_usage('/').used)
        #---------Raw info
        self.total_HD_raw        =(psutil.disk_usage('/').total)
        self.used_HD_raw         =(psutil.disk_usage('/').used)
        self.used_HD_perc        = psutil.disk_usage('/').percent
        #CPU USAGE INFO-------------------------------------------------------------------
        self.cpu_use_perc        = psutil.cpu_percent()
        #SWAP USAGE INFO-------------------------------------------------------------------
        self.swap_used_hum           = humanize.naturalsize(psutil.swap_memory().used)
        self.swap_total_hum          = humanize.naturalsize(psutil.swap_memory().total)
        self.swap_perc               = psutil.swap_memory()[3]
        #---------Raw info
        self.swap_used_raw           = (psutil.swap_memory().used)
        self.swap_total_raw          = (psutil.swap_memory().total)
    def export_to_csv(self):
        print self.hostname
info = Dynamic()

hostname            = info.hostname
iface               = info.iface
ipaddress           = info.ipaddress
total_ram_hum       = info.total_ram_hum
used_ram_hum        = info.used_ram_hum
total_ram_raw       = info.total_ram_raw
used_ram_raw        = info.used_ram_raw
used_ram_perc       = info.used_ram_perc
total_HD_hum        = info.total_HD_hum
used_HD_hum         = info.used_HD_hum
total_HD_raw        = info.total_HD_raw
used_HD_raw         = info.used_HD_raw
used_HD_perc        = info.used_HD_perc
cpu_use_perc        = info.cpu_use_perc
swap_used_hum       = info.swap_used_hum
swap_total_hum      = info.swap_total_hum
swap_perc           = info.swap_perc
swap_used_raw       = info.swap_used_raw
swap_total_raw      = info.swap_total_raw


conn = psycopg2.connect("host='172.31.98.161' dbname='servers' user='seeker'")
cur = conn.cursor() #Create the cursor
#Create a Dictionary to pass the value of each function.
server_info = {'hostname':hostname, 'iface': iface,'ipaddress': ipaddress, 'total_ram_hum': total_ram_hum, 'used_ram_hum': used_ram_hum, 'total_ram_raw': total_ram_raw, 'used_ram_raw':used_ram_raw,'used_ram_perc': used_ram_perc, 'HD_hum': total_HD_hum, 'used_HD_hum': used_HD_hum, 'total_HD_raw': total_HD_raw, 'used_HD_raw':used_HD_raw, 'used_HD_perc': used_HD_perc, 'cpu_use_perc': cpu_use_perc,'swap_used_hum':swap_used_hum, 'swap_total_hum': swap_total_hum, 'swap_perc': swap_perc, 'swap_used_raw': swap_used_raw, 'swap_total_raw': swap_total_raw}
cur.execute("INSERT INTO SERVER_PERF(hostname, iface, ipaddress, total_ram_hum, used_ram_hum, total_ram_raw, used_ram_raw, used_ram_perc, total_HD_hum, used_HD_hum, total_HD_raw,used_HD_raw,used_HD_perc, cpu_use_perc,swap_used_hum, swap_total_hum, swap_perc, swap_used_raw, swap_total_raw) VALUES ('%s','%s','%s','%s','%s','%s','%s','%s','%s','%s','%s','%s','%s','%s','%s','%s','%s','%s','%s')" % (hostname, iface, ipaddress, total_ram_hum, used_ram_hum, total_ram_raw, used_ram_raw,used_ram_perc, total_HD_hum, used_HD_hum, total_HD_raw, used_HD_raw, used_HD_perc, cpu_use_perc, swap_used_hum, swap_total_hum, swap_perc, swap_used_raw, swap_total_raw))
#If this is not present the changes will not get commited.
conn.commit()

The new Database schema is as follows:

pg_dump -U seeker -d servers -s -h 172.31.98.161 > servers_db_schema

--
CREATE DATABASE SERVERS;
CREATE ROLE seeker WITH PASSWORD 'Password!';
ALTER DATABASE SERVERS OWNER TO seeker;
ALTER ROLE seeker WITH LOGIN;
GRANT ALL PRIVILEGES ON DATABASE SERVERS to seeker;
-- PostgreSQL database dump
--

SET statement_timeout = 0;
SET lock_timeout = 0;
SET client_encoding = 'UTF8';
SET standard_conforming_strings = on;
SET check_function_bodies = false;
SET client_min_messages = warning;

--
-- Name: plpgsql; Type: EXTENSION; Schema: -; Owner:
--

CREATE EXTENSION IF NOT EXISTS plpgsql WITH SCHEMA pg_catalog;

--
-- Name: EXTENSION plpgsql; Type: COMMENT; Schema: -; Owner:
--

COMMENT ON EXTENSION plpgsql IS 'PL/pgSQL procedural language';

SET search_path = public, pg_catalog;

SET default_tablespace = '';

SET default_with_oids = false;

--
-- Name: server_perf; Type: TABLE; Schema: public; Owner: seeker; Tablespace:
--

CREATE TABLE server_perf (
hostname text NOT NULL,
iface text,
ipaddress inet NOT NULL,
total_ram_hum text,
used_ram_hum text,
total_ram_raw numeric(30,2),
used_ram_raw numeric(30,2),
used_ram_perc text,
total_hd_hum text,
used_hd_hum text,
total_hd_raw numeric(30,2),
used_hd_raw numeric(30,2),
used_hd_perc text,
cpu_use_perc text,
swap_used_hum text,
swap_total_hum text,
swap_perc text,
swap_used_raw numeric(30,2),
swap_total_raw numeric(30,2),
time_captured timestamp without time zone DEFAULT now()
);

ALTER TABLE server_perf OWNER TO seeker;

--
-- Name: servers; Type: TABLE; Schema: public; Owner: seeker; Tablespace:
--

CREATE TABLE servers (
hostname text NOT NULL,
iface text,
ipaddress inet NOT NULL,
os text,
osrel text,
oskern text,
total_m text,
brand text,
hz text,
cores numeric(4,1),
arch text
);

ALTER TABLE servers OWNER TO seeker;

--
-- Name: pk_hostname; Type: CONSTRAINT; Schema: public; Owner: seeker; Tablespace:
--

ALTER TABLE ONLY servers
ADD CONSTRAINT pk_hostname PRIMARY KEY (hostname);

--
-- Name: server_perf_hostname_fkey; Type: FK CONSTRAINT; Schema: public; Owner: seeker
--

ALTER TABLE ONLY server_perf
ADD CONSTRAINT server_perf_hostname_fkey FOREIGN KEY (hostname) REFERENCES servers(hostname);

--
-- Name: public; Type: ACL; Schema: -; Owner: postgres
--

REVOKE ALL ON SCHEMA public FROM PUBLIC;
REVOKE ALL ON SCHEMA public FROM postgres;
GRANT ALL ON SCHEMA public TO postgres;
GRANT ALL ON SCHEMA public TO PUBLIC;

--
-- PostgreSQL database dump complete
--
Share Button

Python – Get list of processes, their owners and RAM usage from a Linux system.

This script only works on Linux systems. It lists the running PIDs, the owners of the PIDs and the RAM usage of each PID.

import os, sys
from os.path import join,getsize
import humanize
import pwd
import psutil

pids = [int(pid) for pid in os.listdir('/proc') if pid.isdigit()]
for elements in pids:
    p = psutil.Process(elements)
    proc_name  = p.name()
    proc_stat_file = os.stat("/proc/%d" % elements)
    uid = proc_stat_file.st_uid
    username = pwd.getpwuid(uid)[0]
    human_size = humanize.naturalsize(p.memory_info_ex()[0], gnu=True)
    #print  p.memory_info_ex()[0]
    #print username,"\t\t", elements,"\t\t", p.name(), "\t\t", human_size
    print username.ljust(20), elements, p.name(), human_size.rjust(20)

Output Example

root                 17601 cupsd                 3.4M
gmastrokostas        23460 chrome               117.5M
gmastrokostas        24034 pithos                76.2M
Share Button

Python – Get a listing of all subdirectories and their size on a Linux system

The script asks for a directory from the user. It then will dive in all the subdirectories within that directory. It will list all subdirectories and their size. It does this by actually looking for files within the subdirectories and calculating the actual size (not block size) of all the files with in that subdirectory.

#!/usr/bin/python
import humanize
import os, sys
from os.path import join,getsize
import humanize
import pwd

def dir_list():
    list = []
    drct = raw_input(":Enter directory name. Use full path: ")
    for dirpath, dirnames, filenames in os.walk(drct, followlinks=False):
        for loop_dir in dirnames:
            path = os.path.join(dirpath, loop_dir) #Joins the names of directories with the actual path
            list.append(path) # Enters all the directories with the full path to a list

    return list  #Return the whole list so dir_size function can process it

def dir_size():
    
    returned_list = dir_list() #The list returned from the dir_list function

    for loop in returned_list: #Breaks down the list in strings in order for OS.WALK to be able to process it
        total = 0
        for dirpath, dirnames, filenames in os.walk(loop, followlinks= False): #grab the paths from the list. This is the same as asking a user to enter a path

            for f_name in filenames: #Dive into directories to view files
                path = os.path.join(dirpath, f_name) #Join path with filenames for the directory we are trying to find the size
                if os.path.islink(path): 
                    continue
                else:
                    total += os.path.getsize(path)  #Find size of actual files within the directory
                    human_size = humanize.naturalsize(total, gnu=True) 
        print "Size of directory  : ",loop, "is", human_size
        human_size = 0

dir_size()

OutPut Example

FUN 1:Enter directory name. Use full path: /home/gmastrokostas
Size of directory  :  /home/gmastrokostas/.mozilla is 16.1M
Size of directory  :  /home/gmastrokostas/.config is 69.9K
Size of directory  :  /home/gmastrokostas/.cache is 53.0M
Size of directory  :  /home/gmastrokostas/Desktop is 0
Size of directory  :  /home/gmastrokostas/Downloads is 332.9M
Size of directory  :  /home/gmastrokostas/Templates is 0
Size of directory  :  /home/gmastrokostas/Public is 0
Size of directory  :  /home/gmastrokostas/Documents is 218.6M
Size of directory  :  /home/gmastrokostas/Music is 0
Size of directory  :  /home/gmastrokostas/Pictures is 0
Size of directory  :  /home/gmastrokostas/Videos is 0
Size of directory  :  /home/gmastrokostas/.local is 1.2M
Share Button