Backups & Archiving

In Grsync, why does turning on Delete on Destination change the safety profile of your backup, and what is the default?

A colleague asks which Linux backup tool is objectively the best. Why is there no single right answer, and what factors actually drive the choice?

Someone runs Deja Dup for the first time expecting only their changed files to be copied. What actually happens on that first run versus later runs, and how can restored data be read back?

An admin sets up Deja Dup to protect a whole server including /etc and installed system files. Why is that the wrong tool for the job?

Deja Dup and Back In Time both do incremental backups, but restoring a single old file behaves very differently. Where does the duplicity-based approach hurt?

When configuring what Deja Dup should skip, which folders are ignored by default, and which large items are worth adding manually?

Back In Time is described as fast and simple to restore from. What mechanism gives it those properties, and what filesystem requirement does that impose?

What practical restore advantage does Back In Time have over Deja Dup even when Back In Time itself is unavailable?

Back In Time's Intelligent Delete is meant to keep the backup volume from growing forever. Which backups does it guarantee to keep, and which does it promise never to touch?

A user runs Back In Time normally and assumes /etc and other system files are protected. What did they actually back up, and how do you include system files?

Once Back In Time is configured, how do its scheduled backups actually run, and does the GUI need to stay open?

Grsync markets itself as a backup tool but is fundamentally a GUI for rsync. What does that make it good at, and what two limitations follow?

Borg Backup is often suggested for backing up straight to Amazon S3. Why is that a mismatch, and what targets does Borg actually support?

Borg claims strong space efficiency even when many files repeat across archives. What technique makes that possible, and what constraint does it put on restores?

Before your first Borg backup you have to initialize a repository. What command does that, and where does the encryption key end up?

How do you create a named Borg backup of a directory, and how is the archive name expressed in the command?

Borg's --stats output shows three size numbers. What do Original, Compressed, and Deduplicated size each tell you, and which one reflects the real cost of a repeat backup?

In Borg, what is the difference in output between listing a repository and listing a specific archive?

You run borg extract to recover a backup into a directory that already has files. What does Borg do to those existing files, and how do you restore just a few files instead of everything?

To trim old Borg backups you could use borg delete, but borg prune is usually preferred. What does prune do that delete does not?

In an automated Borg script, how do you avoid hardcoding the repo path and passphrase into every command, and how do you give each backup a date-based name?

How do you point Borg at a remote repository over SSH, and what must be true of the remote machine?

Why does Borg default to lz4 compression rather than gzip, and what surprising limit applies when you switch compression on an existing repository?

gzip is the default compressor for text, but the book warns against using it for certain files. Which algorithm does it use, and what data is it a poor fit for?

Among gzip, bzip2, xz, and lzop, how do they rank on compression ratio versus speed, and when would you deliberately pick the one that compresses worst?

You want to stream a logical volume over the network with minimal CPU overhead. What compressor fits, and roughly what is the cost compared to a raw copy?

In the tar command tar -czf archive.tgz dir/, what does each of the three option letters mean, and what extensions signal this format?

With a gzip-compressed tar archive, how do you list its contents versus extract it, and how do you pull out only files matching a pattern?

You have a tar workflow built around gzip but need bzip2 compression instead. What single change to the options switches the codec?

When you need to hand an archive to a Windows user, why choose zip over tar, and what are the create/inspect/extract commands?

rsync is described as synchronizing directory trees rather than plain copying. What does it do differently on the first pass versus later runs, and why are its network transfers encrypted by default?

rsync accepts several remote-location syntaxes. How do host:dir, user@host:dir, and rsync://user@host/dir differ in what they connect to?

The rsync -a (archive) flag is a shorthand. Which behaviors does it bundle, and what does that preserve that a plain recursive copy would not?

An admin adds --delete to their rsync backup so it stays a clean mirror. Why is this the classic way to lose data, and what is rsync's default without it?

You want an rsync backup to run unattended from cron. How do you avoid the interactive password prompt, and what security precaution does the book recommend?

Running a dedicated rsync server (daemon) instead of using SSH pulls in a network dependency. What port must be open, and when is running a server actually worthwhile?

Someone tries to rsync backups directly to a typical cloud storage service and it fails. Why, and what tool does the book suggest for cloud targets?

rdiff-backup is pitched as an alternative to rsync that behaves like Time Machine. What does it do with old versions of changed files that plain rsync does not?

In an rdiff-backup target directory, how is the most recent backup stored, and where do the old versions live?

How do you restore from rdiff-backup, and what time syntax lets you grab either the latest state or a state from N days ago?

An rdiff-backup directory keeps growing. How do you prune it by age versus by number of retained versions?

For remote backups, what installation and syntax difference sets rdiff-backup apart from rsync?

rsnapshot and rdiff-backup both give versioned incremental backups on top of rsync. What does rsnapshot use to reference unchanged files, and what feature does it give up in exchange?

The book calls rsnapshot's design the opposite of most backup tools. What direction does it move data, and what problem is that shape built for?

rsnapshot's config file is notorious for silent syntax errors. What are the two formatting rules in /etc/rsnapshot.conf that trip people up?

Which three parameters in rsnapshot.conf do you actually need to understand to configure it, and what is the default backup location?

In rsnapshot, what do interval settings like hourly 6, daily 7, weekly 4 actually control, and why does adding more intervals cost so much storage?

Given rsnapshot's default snapshot_root, where would you find the most recent hourly backup of the local /etc directory?

You enabled a monthly rsnapshot job in /etc/cron.d/rsnapshot but it errors out. What consistency rule between the cron file and the config file did you likely miss?

You drop a backup script into /etc/cron.daily but it never runs. What naming and permission requirements does run-parts impose, and what is the alternative /etc/cron.d form?

A cron rsync backup to an external disk runs even when the disk is unplugged, and the source ends up mirrored to nothing. What one check makes the script safe?

A daily tar script names files with date +%d and date +%m. How does this yield roughly 43 rotating versions, and what is the retention pattern?

Why is Amazon S3 especially valued as backup storage, and what unusual aspect of its pricing should shape how you write backup scripts?

Two rules govern naming and securing an S3 bucket for backups. What are they, and what real-world failure motivates the second?

What do the core aws s3 subcommands ls, cp, and sync each do, and what does sync leave behind by default?

An admin expects to mkdir and cd inside an S3 bucket and is confused when names like d1/d2/file still appear as folders. How does S3 actually organize objects?

An aws backup command works fine at the shell but silently fails under cron on some distributions. What environment difference causes this, and how is it fixed?

Transfers to S3 are already encrypted in transit, yet the book insists you still encrypt files yourself before upload. Why, and what workflow pattern is recommended?

For encrypting backups before cloud upload, how do you generate a strong symmetric key and apply it, and what stronger approach does the book recommend for higher security?

A common cloud DB-backup design splits the work across two cron scripts. What does each stage do, and how does the filename provide rotation?

Why is a backup procedure that runs successfully every night still not trustworthy until you do one more thing?

Before the aws CLI can touch S3, what IAM setup and configuration step are required, and where do the credentials end up on disk?