## Simple solutions

Solène Rapenne
BitreichCON 2019

## About me

+ Sysadmin since 2011

+ Managing FreeBSD hosts for a living

+ Member of OpenBSD project since 2018

## SIMPLE SOLUTIONS

+ Simple tools

+ Makefile

+ Drist

+ Crontab

+ Tips

## Simple tools

+ know your shell (tests, loops, special variables, limitations)

+ find / awk / grep

+ reuse data

+ have fun and optimize later

+ learn AWK

## Simple tools examples

+ using curl+AWK for making a RSS file from
  https://www.openbsd.org/faq/current.html

+ bitreich lawn status

## Have fun example #1

Building a small gopher client

1. Fetch data

    printf "/\n" | nc bitreich.org

2. Only display actual text

    | cut -d '	' -f 1

3. Display "i" fields without i and indent others types

   awk '{ if(substr($0,0,1)=="i")
        { print substr($0,2) }
        else { printf "\t%s\n", $0 }}'

## Have fun example #2 part 1

Extract article titles from my blog

1. Fetch data

    curl https://dataswamp.org/~solene/index.html

Titles are within a a tag and link is relative and starts with a date.
Output is html

2. Filter on the previous pattern detected

    | grep "<a href=\20"

Because all articles were written in current century this is easy.

Output looks like:
<h1><a href="2016-06-08-git.html">Git cheat sheet</a></h1>
<h1><a href="2016-06-07-13.html">How to send html signature in mu4e</a></h1>
<h1><a href="2016-06-06-stumpwm.html">My Stumpwm config on OpenBSD</a></h1>
<h1><a href="2016-05-31-mbuffer.html">Port of the week: mbuffer</a></h1>
<h1><a href="2016-05-25-12.html">FreeBSD 11 and Perc H720P Mini raid controller</a></h1>


## Have fun example #2 part 2

3. Separate content from tags

It's easier to work on lines

    | tr '<>' '\n'

Output looks like:
h1

a href="2016-05-02-3.html"
How to add a route through a specific interface on FreeBSD 10
/a

/h1


4. Get line after a "a href"

    | awk '/^a href/ { getline; print}'

Output:
My Stumpwm config on OpenBSD
Port of the week: mbuffer
FreeBSD 11 and Perc H720P Mini raid controller
Port of the week: rdesktop


## Learn AWK

+ available everywhere

+ sed / grep / cut / tr all in one

+ Print penultiem field

    awk '{ print $(NF-1) }'

+ Reorder fields

    awk '{ print $3" "$2 }'

+ Extensible (contrary to a combination of sed / grep / cut)

+ Avoid writing it as one liner!

## Makefile

+ Idea from bitreich

+ Standard interface

+ Use it everywhere

+ Stop on error

+ Replacement for scripts?

## Drist

+ simple of use

+ super fast

+ but not perfect

+ awesome with makefile!

## Drist tips

+ Install package if not present

    ls /usr/local/bin/nginx || pkg install -y nginx

+ Add a line at the end of line (ONCE)

    grep ^md99 /etc/fstab || \
    ( echo "md99 none swap sw,file=/swap.disk 0 0" >> /etc/fstab )

+ Add a line at a specific place

    FILE=/usr/local/etc/lighttpd/modules.conf
    PATTERN=\"mod_openssl\"
    grep "$PATTERN" "${FILE}" || ( printf '%s\n' 1a ' "${PATTERN},' . w | ed $FILE )

+ Fetch remote data into the drist directory

    rsync -aRv remote:/home/solene/.ssh/ files/

## Crontab

+ available everywhere

+ run your scripts while having fun

+ have some kind of output (mail or file)

+ I solved it lacks of automation with awk

gopher://dataswamp.org/0/~solene/article-crontab-manipulation.txt
https://dataswamp.org/~solene/2018-05-31-crontab-manipulation.html

    This includes a text file within defined tags (in comment) within crontab
    and replace the content on each run.


## Tips

+ Write everything you do

Store most of your work, you may need it later

+ Track file history with a version control system

Keep track of every versions of your files.

+ Handling Failing script

    command || ( echo failure | mail solene )

+ Checking changes between runs

    command_something > new
    diff -u old new > diff
    if [ -s diff ]; then do something it changed! ; fi
    mv new old