Script para auto-arquivamento das configurações de elementos de rede

Certo dia, no trabalho, estava pesquisando no Google por um script que automatizasse a tarefa de extrair as configurações dos elementos de rede de tempos em tempos e as armazenasse em um servidor para consultas futuras. Depois de entrar em alguns sites, acabei encontrando exatamente o que eu procurava no site LinuxQuestions.org. Testei o script em nossa rede e funcionou perfeitamente. Compartilho aqui, com vocês, o dito-cujo:

#!/bin/bash
+++++++++++++++++++++++++++++
# Name: cisco_tftp_backup
# Author: Steve Cowles 
#
# Revision: Created 10/1/2005 SWC
# Revision: 03/26/06 - SWC
# Added capability to save config data by date.
#
# Description: Shell Script to backup cisco startup-config files
# using tftp and store in pre-defined directory structure
#
# Directory structure for script is:
# base_dir /backup/cisco/{date}
# device type /router
# hostname /r1
# filename startup-config
# hostname /r2
# filename startup-config
# device type /switch
# hostname /s1
# filename startup-config
# hostname /s2
# filename startup-config
#
# 10.1.100.201 is the IP address of the system running this script
# nvram:startup-config is the only file allowed to be copied
+++++++++++++++++++++++++++++

+++++++++++++++++++++++++++++
# Error handlers
error_no_tftp_exec ()
{
  echo "ERROR - Filename '$1' does not exist"
  exit
}

error_no_backup ()
{
  echo "ERROR - Unable to backup Host '$1'"
}

error_zero_length ()
{
  echo "ERROR - Hostname '$1' backup has a zero size"
}
+++++++++++++++++++++++++++++

# If tftp executable does not exist... then exit
CMD=tftp ; TFTP=`which ${CMD} 2>/dev/null`
[ ! -x "${TFTP}" ] && error_no_tftp_exec ${CMD}

+++++++++++++++++++++++++++++
# Variable Section, edit below to meet requirements
+++++++++++++++++++++++++++++

# Define/load a variable to store the device information
# of all routers/switches to backup using tftp.
#
# Values are separated by colons (:)
# Value 1 = Device Type (sub-dir created)
# Value 2 = Device Hostname (sub-dir created)
# Value 3 = Device IP address
DEVICES="
router:allar1:10.100.12.1
switch:allsw1:10.100.12.2
router:lewar1:10.100.16.1
switch:lewsw1:10.100.16.2
"

# Should an error occur during execution, e-mail
# errors to following person. i.e. If run as cronjob
MAILTO="[email protected]"

# Define the base directory where you want to store
# files retreived from all devices. All subordinate
# directories will created relative to this base
BASELOG=/var/log/cisco_configs
BASEDIR=${BASELOG}/`date +%m.%d.%y`

# Define the filename to (get) from device using tftp
# See note 1 above
FILENAME=startup-config

+++++++++++++++++++++++++++++
# END Variable Section, No servicable parts below
+++++++++++++++++++++++++++++

+++++++++++++++++++++++++++++
# Begin Executable Section (do NOT edit below)
+++++++++++++++++++++++++++++

# If base directory does not exist, create it
[ ! -d ${BASEDIR} ] && mkdir ${BASEDIR}

# Update the symbolic link to point to the current 
# BASEDIR directory
if [ -d ${BASELOG} ] ; then
  # Router link
  rm ${BASELOG}/router
  ln -s ${BASEDIR}/router ${BASELOG}/router

# Switch link
  rm ${BASELOG}/switch
  ln -s ${BASEDIR}/switch ${BASELOG}/switch
fi

+++++++++++++++++++++++++++++
# Setup loop for each device listed in $DEVICES variable
+++++++++++++++++++++++++++++
for device in $DEVICES ; do
  # Separate DEVICE TYPE/HOSTNAME/IP into separate varaibles
  DEVICE=`echo ${device} | cut -d ':' -f 1`
  HOSTNAME=`echo ${device} | cut -d ':' -f 2`
  IP=`echo ${device} | cut -d ':' -f 3`

# Set and Create the sub-directories to store files
  SUBDIR=${BASEDIR}/${DEVICE}
  [ ! -d ${SUBDIR} ] && mkdir ${SUBDIR}
  [ ! -d ${SUBDIR}/${HOSTNAME} ] && mkdir ${SUBDIR}/${HOSTNAME}

# Set the fullpath to store file retreived during tftp
  FULLPATH=${SUBDIR}/${HOSTNAME}/${FILENAME}

# tftp $FILENAME from device/IP to directory/FULLPATH
  ${TFTP} $IP -c get ${FILENAME} ${FULLPATH} >/dev/null 2>&1 ||
  error_no_backup ${HOSTNAME}

# Since tftp return values don't seem to include zero length
  # gets, test for this condition
  [ ! -s ${FULLPATH} ] && error_zero_length ${HOSTNAME}
done

Destaquei em amarelo onde a lista com os elementos e seus IPs deve ser incluída.
PS: Não se esqueça que é preciso liberar, nos elementos de rede Cisco, o acesso do servidor de backup. Isso pode ser feito via exemplo abaixo:

access-list 55 remark PERMIT hosts requesting TFTP access
access-list 55 permit host 10.1.100.201
tftp-server nvram:startup-config 55

Onde 10.1.100.201 é o endereço IP do servidor. No exemplo, o script vai buscar a startup-config. Mas ele pode ser facilmente alterado para buscar a running-config ou outro arquivo.

Abaixo, um exemplo do que ele faz, usando o nosso servidor de teste:

[marcoaf@server tftp]$ ls
08.17.10 08.21.10 08.25.10 08.29.10 09.02.10 09.06.10
08.18.10 08.22.10 08.26.10 08.30.10 09.03.10 09.07.10
08.19.10 08.23.10 08.27.10 08.31.10 09.04.10 09.08.10
08.20.10 08.24.10 08.28.10 09.01.10 09.05.10

[marcoaf@server tftp]$ cd 09.05.10/
[marcoaf@server 09.05.10]$ ls
acs firewall router switch

[marcoaf@server 09.05.10]$ cd router/
[marcoaf@server router]$ ls
ROTEADOR01 ROTEADOR02 ROTEADOR03 ROTEADOR04

[marcoaf@server router]$ cd ROTEADOR02
[marcoaf@server ROTEADOR02]$ ls
running-config

Tudo organizado por data e tipo de elemento. Para que o script funcione, você deve torná-lo executável no Linux e incluí-lo no CRONTAB para rodar no intervalo que você quiser.