Need help understanding a back-up script

Hello everyone!

I had a container with a DB crap itself yesterday so I’m trying to speed up my learning to back up stuff.

I came across a script that taught me how to back-up a containerized postgres db at given intervals and it works. I managed to create db dumps and restore them. I’ve documented everything and now my whole docker-compose/env etc are on git control.

There’s one part of the script I don’t decypher but I’d like to maybe change it. It is about the number of back-up copies.

Here’s the line from the tutorial: ls -1 /backup/*.dump | head -n -2 | xargs rm -f

Can someone explain to me what this line does? I’d like to keep maybe 3 copies just in case the auto-backup backs up a rotten one.

Thanks!

Full code below:


<span style="color:#323232;">backup:
</span><span style="color:#323232;">    image: postgres:13
</span><span style="color:#323232;">    depends_on:
</span><span style="color:#323232;">      - db_recipes
</span><span style="color:#323232;">    volumes:
</span><span style="color:#323232;">      - ./backup:/backup
</span><span style="color:#323232;">    command: >
</span><span style="color:#323232;">      bash -c "while true; do
</span><span style="color:#323232;">        PGPASSWORD=$$POSTGRES_PASSWORD pg_dump -h db-postgresql -U $$POSTGRES_USER -Fc $$POSTGRES_DB > /backup/$$(date +%Y-%m-%d-%H-%M-%S).dump
</span><span style="color:#323232;">        echo ""Backup done at $$(date +%Y-%m-%d_%H:%M:%S)""
</span><span style="color:#323232;">        ls -1 /backup/*.dump | head -n -2 | xargs rm -f
</span><span style="color:#323232;">        sleep 86400
</span><span style="color:#323232;">      done"
</span>
un_ax,

If you want to get more in depth, I’ve been using this container:

github.com/jareware/docker-volume-backup

It can be setup in the same compose or in it’s own, and it supports pre/post commands if you want to dump a db or stop a container before backup.

Additionally, Setting a post backup command like in their docs:

POST_BACKUP_COMMAND: “docker run --rm -e DRY_RUN=false -e DAILY=3 -e WEEKLY=1 -e MONTHLY=1 -v /backup:/archive ghcr.io/jan-brinkmann/docker-rotate-backups”

Lets you specify the number of backups retained per period, E.G. 3 daily, 1 weekly, 1 monthly.

You could also mix and match.

Doomdoxrulz,

The first command (ls -1 /backup/*.dump) just creates a list of files in the backup folder that have the extension .dump. the output of the prior command is then sent to the next command (head -n -2) this cuts the list down to everything except the last 2 items in the list this is then sent to the final command which takes the list and runs the final (rm -f) command with the items in the list as the targets to delete.

heres a solution based on this post stackoverflow.com/…/delete-all-but-the-most-recen…

ls -tp /backup/*.dump | grep -v ‘/$’ | tail -n +4 | tr ‘\n’ ‘\0’ | xargs -0 rm -f

There is an explanation on that post that explains it in better detail but in simple terms it deletes all files but the most recent 3 files in the directory that have the .dump extension

tburkhol,

Others have explained the line.

Worth noting that not all implementations of head accept negative line counts (i.e. last n lines), and you might substitute tail.

i.e.: ls -1 /backup/*.dump | tail -2 | xargs rm -f

klay,

Won’t this delete the two newest files, as opposed to everything except the two newest files?

doeknius_gloek,
@doeknius_gloek@feddit.de avatar

Yeah, tail would be the more obvious choice instead of negating head.

Fuck, I need coffee. @klay is right (again).

doeknius_gloek, (edited )
@doeknius_gloek@feddit.de avatar

This line seems to list all dumps and then deletes all but the two most recent ones.

In detail:

  • ls -1 /backup/*.dump lists all files ending with .dump alphabetically inside the /backup directory
  • head -n -2 returns all filenames except the two most recent ones from the end of the list
  • xargs rm -f passes the filenames to rm -f to delete them

Take a look at explainshell.com.

klay,

I just looked up the man page, and actually head -n -2 means “everything up to but not including the last two lines”, so this should always leave two files remaining.

doeknius_gloek,
@doeknius_gloek@feddit.de avatar

You’re right, I edited my comment. Thanks!

mnmalst,

Your xargs comment is still wrong tho. It deletes ALL but the most recent two files.

doeknius_gloek,
@doeknius_gloek@feddit.de avatar

Fixed, thanks.

smileyhead,

Backups are created to /backup directory and are ended with .dump file extention.

ls -1 is listing all those files chronologically, -1 is to keep one file per one line.

head -n -2 is getting lines from the top to the last two at bottom.

xargs rm -f is calling rm -f on every line of the input.

| is pipe symbol, that gets output from command before and gives it to command after

So TLDR it’s removing all backups except the last 2 ones.

klay, (edited )

Ah! This is a shell pipe! It’s composing several smaller commands together, cool stuff.

  • ls -1 is the grep-friendly version of ls, it prints one entry per line, like a shopping list.
  • head takes a set number of entries from the head of a list, in this case 2 items. negative two, meaning “all but the last two.”
  • xargs takes the incoming pipe and converts it into extra arguments, in this case applying those arguments to rm.

So, combined, this says “list all the .dump files, pick the first two, all but the last two, and delete them.” Presumably the first are the oldest ones and the last are the newest, if the .dump files are named chronologically.

macallik,

Great response 👍🏾

  • All
  • Subscribed
  • Moderated
  • Favorites
  • •
  • random
  • uselessserver093
  • Food
  • aaaaaaacccccccce
  • [email protected]
  • test
  • CafeMeta
  • testmag
  • MUD
  • RhythmGameZone
  • RSS
  • dabs
  • Socialism
  • KbinCafe
  • TheResearchGuardian
  • oklahoma
  • feritale
  • SuperSentai
  • KamenRider
  • All magazines
    •