If you have a bash script running with a very long loop (or infinite one) and starting multiple background processes, you will notice that when you stop the main process, all the child processes will continue to live until you restart the machine or you kill them one by one.

So that you can easily stop all the sub-processes when you stop the main process, you just need to use the trap function to capture the event and stop all the jobs in the background.

Here’s an example of code, executing 2 infinite functions:

#!/bin/bash

# Get number of seconds of uptime
func_1() { 
    while true
    do
        echo "Press Ctrl-C to stop me even if I'm a child..."
        sleep 10
    done
}

func_2() {
    while true
    do
        echo `date`
        sleep 3
    done
}

#####################
### MAIN PROGRAM ####
#####################

declare -a pids
declare -a functions=(func_1 func_2)

# Parse list of devices
i=0
for func in "${functions[@]}"
do
    ${func} &
    pids[${i}]=$!
    i=$((i+1))
done

trap "trap - SIGTERM && kill -- -$$" SIGINT SIGTERM EXIT

# wait for all pids
for pid in ${pids[*]}; do
    wait $pid
done

Output of the script until we hit Ctrl-C:

# bash stop_proc.sh
Press Ctrl-C to stop me even if I'm a child…
Tue Nov 14 22:28:55 UTC 2022
Tue Nov 14 22:28:58 UTC 2022
Tue Nov 14 22:29:01 UTC 2022
Tue Nov 14 2:29:04 UTC 2022
Press Ctrl-C to stop me even if I'm a child…
Tue Nov 14 22:29:07 UTC 2022
Tue Nov 14 22:29:10 UTC 2022
^CTerminated

Sometimes you might need to download some pip packages offline on a device with no Internet access. This is entirely feasible through multiple ways.

Install a single package:

You first need to download that package on a device with Internet access (using pip package as an example)

wget https://files.pythonhosted.org/packages/6a/df/a6ef77a6574781a668791419ffe366c8acd1c3cf4709d210cb53cd5ce1c2/pip-22.0.3-py3-none-any.whl

Then you push that file to the device you need to install the package on, and you can execute:

pip install pip-22.0.3-py3-none-any.whl

Install a list of requirements:

You first need to download the list of packages needed from the requirements.txt file:

pip download -r requirements.txt

Then, push all the downloaded files to the server, and install those requirements on the server, using the downloaded packages:

pip install --no-index --find-links /path/to/download/dir/ -r requirements.txt

If you get the following error when trying to execute a sudo command, despite the fact that you configured your sudoers or sudoers.d config file properly:

sudo: pam_unix(sudo:auth): conversation failed
sudo: pam_unix(sudo:auth): auth could not identify password for [user]

Then, you might need to have a look at the PAM configuration on your device.
Ensure that you have this line enabled in your /etc/pam.d/sudo file:

# cat /etc/pam.d/sudo
#%PAM-1.0
# The line below should fix the sudo:auth could not identify password
auth       sufficient   pam_permit.so

# The lines below are the default config
auth       include      system-auth
account    include      system-auth
password   include      system-auth
session    optional     pam_keyinit.so revoke

session    include      system-auth

Once you changed this file, no need to restart anything – just retry your command and it should work!

If you want to filter some specific logs and redirect them to another server or to another file, you can use the module provided with rsyslog called “omprog”.

Thanks to that module, you can just ask rsyslog to execute a script on every logs you have from a specific application and process it with your own rules.

Let’s go with a simple example, I want to filter my logs from postfix and only extract few specific information.

Below is the configuration I will now push to /etc/rsyslog.d/rsyslog_postfix.conf in order to:

1. Redirect the current log to /var/log/maillog as a default path (to keep the raw log)
2. Process every log line with my python script in order to extract only some specific informations that I will write in /var/log/rsyslog_postfix.log

module(load="omprog")

mail.* {
    action(type="omfile" 
        file="/var/log/maillog")
    action(type="omprog"
        name="rsyslog_postfix"
        binary="/usr/bin/python3 /opt/rsyslog_postfix.py"
        queue.type="LinkedList"
        queue.size="20000"
        queue.saveOnShutdown="on"
        queue.workerThreads="4"
        queue.workerThreadMinimumMessages="5000"
        action.resumeInterval="5"
        output="/var/log/rsyslog_postfix.log")
    stop
}

Rsyslog script

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
# python_version  :3.6
#===========================

import sys
import traceback
import datetime

if __name__ == '__main__':
    data = {}
    while True:
        try:
            line = sys.stdin.readline()
            msg = line.strip()
            if msg != "":
                data.clear()
                timestamp = msg.split(' ')[0]
                proc= msg.split(' ')[2].split('[')[0]
                pid = msg.split(' ')[2].split('[')[1].split(']')[0]
                log = msg.split(' ', 3)[3]
                if proc.startswith("postfix/"):
                    data = { 
                        "Timestamp": timestamp,
                        "Process": proc,
                        "PID": pid,
                    }
                    data["PostfixID"] = log.split(' ')[0].split(':')[0]
                print(data)
        except Exception as e:
            err = traceback.format_exc()
            print(err)

Now, I just need to restart rsyslogd process and the logs will start to be processed as expected, and you should get the raw logs in /var/log/maillog and the filtered ones in /var/log/rsyslog_postfix.log

systemctl restart rsyslog

FIPS (Federal Information Processing Standard) can be enabled (by default or not) on linux kernels to enable the FIPS kernel cryptographic features.
But in some case, this can also lead to some issues with openssl, or any cryptographic tool that you can use within any code.

You can check if FIPS is enabled with that command:

# cat /proc/sys/crypto/fips_enabled
1

If you need to turn this feature off, you will have to first remove any dracut-fips package that you have installed:

# yum -y remove dracut-fips* 

Then, take a backup of the FIPS initramfs and recreate a new file:

# cp -p /boot/initramfs-$(uname -r).img /opt/initramfs-$(uname -r).backup
# dracut -f 

Once the file creation is complete, update your GRUB configuration to disable fips flag and rebuild grub configuration

# perl -pi -e 's/fips=1/fips=0/g' /etc/default/grub 
# grub2-mkconfig -o /boot/grub2/grub.cfg
# grub2-mkconfig -o /boot/efi/EFI/centos/grub.cfg

Now, reboot the server and if you check FIPS status again, it should return a value of 0 indicating that FIPS is disabled:

# cat /proc/sys/crypto/fips_enabled
0

When analyzing the output of a bash script, it can be useful to prepend timestamp before each line to see how long it’s taking to do a specific action.

For this, you can use the ‘ts‘ command:

# echo -e "this\nis\na\ntest" | ts '[%Y-%m-%d %H:%M:%S]'
[2019-05-13 09:14:28] this
[2019-05-13 09:14:28] is
[2019-05-13 09:14:28] a
[2019-05-13 09:14:28] test

If the command ‘ts‘ is not available, you might need to install the package moreutils.

Seeing that error in HTTPd/Apache logs when trying to GET some pages?

(13)Permission denied: [client xxxxx]: access to /html/myfile.html denied because search permissions are missing on a component of the path

This is because of some SELinux policies blocking the calls. If you don’t want to turn SELinux off, a simple workaround is:

chcon -R --type=httpd_sys_rw_content_t /html/

When updating or installing a package on a Debian-based distribution, you can face that alert:

W: GPG error: http://ppa.launchpad.net trusty Release: The following signatures couldn't be verified because the public key is not available: NO_PUBKEY C2518248EEA14886

This just means that a signing GPG key is missing or expired on your system. The error is giving you the ID of the missing key that you can use for looking for it on the Ubuntu website:

http://keyserver.ubuntu.com:11371/pks/lookup?op=get&search=0xC2518248EEA14886

Copy the key that you just got from the website above into a file – mykey.txt, for example – and import it with the apt-key command:

apt-key add mykey.txt

Try to install/update the package again, and the error should not appear anymore.

The PKCS12 format is replacing the old PFX format from Microsoft. This format will allow storage of X.509 private keys and the associated public certificates in a single encrypted file.

So you can extract the key and the certificate in a single common PEM file, you can use this openssl command:

openssl pkcs12 -in myCertificate.pfx -out myCertificateAndMyKey.pem

If you want to extract the key and the certificate independently, you can also use the options nocerts/nokeys along with openssl, to extract only one part:

openssl pkcs12 -in myCertificate.pfx -nocerts -out myCertificate.key
openssl pkcs12 -in myCertificate.pfx -nokeys -out myCertificate.pem

When doing update on a Debian/Ubuntu server, if a package is trying to modify a config file, it’s asking for a manual choice. If you’re trying to try to automate the process, you’d probably like to get rid of that ask and keep the current config file by default.

For that, you can use that script:

#!/bin/bash
apt-get update
listUpgrades=`apt list --upgradable |grep upgradable |cut -d/ -f1`
execUpgrades="DEBIAN_FRONTEND=noninteractive apt-get --yes --assume-yes -o DPkg::options::=\"--force-confdef\" -o DPkg::options::=\"--force-confold\" --only-upgrade install "$listUpgrades
eval $execUpgrades

The multiple options will enforce the update to keep current config and not overwrite it with the package default one.