0:03

So in a change to our regularly scheduled

0:06

programming, I thought we'd all talk about

0:08

the same topic this week, but from different

0:11

angles. Okay. Yeah.

0:13

Yeah. Cool. Backups.

0:16

You should do them. You really should do them is the overriding message

0:18

you should get from this. And I want to talk about how I do

0:21

backups and then you guys can talk about

0:23

yours. So I'll start with mine.

0:26

Most of my systems are Linux laptops,

0:29

desktops, servers, they're

0:31

all Linux. And so

0:33

they pretty much all reside in my house

0:36

or are remotely accessible via

0:39

SSH. So I've got a few servers out

0:41

there in the cloud. I only have one

0:43

location. I'm not backing up from lots

0:45

of places other than out there on the internet.

0:48

And so for many years, I've used the

0:50

same tool for doing backups and it's

0:52

called our snapshot. And

0:56

it's been around a long time.

0:57

It's a little bit quirky

1:00

and it's a little bit weird

1:02

to set up. But once it's set up,

1:04

it's pretty much fire and forget.

1:07

So I have a server upstairs in the loft

1:10

at home

1:11

and that's running our snapshot.

1:13

So it has a regular

1:15

cron job to run our snapshot.

1:18

In fact, it has four cron jobs to run our

1:20

snapshot. And what it does is

1:23

the configuration for our snapshot tells

1:25

it which machines to back up remotely. And

1:28

it does that over SSH as root.

1:31

So it has the SSH keys all set

1:33

up so that machine can SSH

1:36

to any system as

1:38

root and access

1:40

the file system of that remote system. And then it

1:42

copies that all the files from that remote

1:45

system onto the server at home.

1:47

And it does that on a regular basis

1:49

for multiple machines. It wakes

1:52

up

1:52

every few hours and backs

1:55

up my laptop, my desktop,

1:57

my web server, a game

1:59

server.

1:59

of run running and all kinds of other machines.

2:02

And the neat thing about

2:04

R snapshot is it uses R sync over

2:07

SSH and it only

2:10

backs up changed or new files. So

2:14

it's not doing a full backup every single time. It's

2:17

the first backup takes a long time because it backs

2:19

up everything that you specify and you don't have to back

2:21

up the whole system. You can just back up certain folders.

2:24

So I only backup certain important folders.

2:27

But the important part is it

2:29

runs on a regular basis and

2:32

backs up all the files on all the systems.

2:34

And it works.

2:36

And do you just get one backup of

2:38

whatever the latest state is or

2:41

does it do some sort of time

2:43

window where you can go back and say, I need

2:46

to know what it was yesterday or something like that?

2:48

Yeah, it's configurable. Right. So

2:51

the way I have it set up

2:52

is that it runs every six hours

2:56

and it keeps a certain number of those

2:58

backups.

2:59

When it gets to the limit of how

3:01

ever many you set, let's say it's six,

3:04

it renames the

3:06

oldest one

3:07

to another name. And they're called

3:10

alpha, beta, delta, gamma.

3:13

And so the alpha backups, I've got six

3:15

alpha backups. And then when it starts the next

3:18

one and it sees there are already six that have been

3:20

done over the last six backups,

3:23

it takes the oldest one and renames it beta.zero.

3:27

And then away we go again and it does another

3:29

backup.

3:30

And then the next time it runs, beta

3:32

zero gets renamed beta one until

3:34

we have enough beta backups. And

3:36

then the oldest beta backup gets renamed delta

3:39

and then the oldest delta one gets renamed gamma.

3:42

The net result of that is I have

3:44

backups going quite far

3:46

back, but I only keep

3:49

seven of the most recent ones.

3:52

And then the others are daily,

3:54

the beta ones are daily. And then the

3:57

older ones are weekly. And then the older ones on that

3:59

are monthly.

3:59

So I effectively have hourly,

4:02

daily, weekly, monthly, roughly,

4:06

but because I don't back up every hour, it's not actually

4:08

hourly, it's every six hours

4:10

and then every day and then every week and then every month. So

4:13

I go back like six months, maybe a year

4:16

of backups,

4:17

but not every single backup. It's not

4:19

like ZFS snapshots, it's

4:22

a delta taken at a snapshot in time

4:25

at some point over the last year. I've

4:28

been doing this for years on all my systems and

4:31

it has saved my bacon a bunch of times

4:33

where I've aggressively

4:35

deleted files off of a laptop and then wanted

4:38

to go and find them or needed

4:40

to go and find some configuration file

4:43

from a year ago because I've had to redeploy

4:45

something or something like that.

4:47

There are certainly better ways to

4:49

do it. It's not the fastest backup in the

4:51

world because the way

4:53

it copies the files from Alpha0

4:56

to Alpha1 and then Alpha1 to Alpha2,

4:58

it takes a lot of time to do that copying. It's basically

5:01

CP, a folder to another folder.

5:03

It does save space by using hard links so

5:06

you don't end up with lots of duplicate copies

5:08

of the same files. But

5:10

if you modify a lot of files, then a lot of files

5:12

get changed and a lot

5:14

of files get backed up. But I basically

5:17

don't look at it very often because whenever

5:19

I look at the log file, it's just working and

5:22

it has done for years. And so

5:24

it's good old reliable R snapshot

5:27

and I'm quite happy with it. And when I've never needed

5:30

to get files from it, they've been there and

5:32

that's kept

5:33

me very happy. And I think that's

5:35

a critical piece, right? Because people

5:37

talk about making backups and

5:40

I've spoken to people about the rigor

5:42

with which they take backups. And I was like,

5:45

what's the restore process like? And they're like, no,

5:47

I've never really tested it. Yeah,

5:51

well, you should probably do that because you

5:53

may just have a load of garbage. Yeah.

5:56

And I've restored like entire

5:59

folders full of.

5:59

music and folders

6:02

full of PDFs and stuff. It's not just individual

6:04

files, but it's all there, going

6:06

back over a year.

6:09

And I've recently been getting

6:11

rid of some of my old backups because I just don't

6:13

need them anymore. I'm on my fourth

6:16

generation of laptop that I've

6:18

been using our snapshot on with this

6:20

server, and it's keeping all

6:22

those old laptop snapshots there.

6:25

I call them snapshots, they're backups really.

6:28

And so every time it's copying alpha 0 to alpha 1

6:30

and alpha 1 to alpha 2, it's copying not

6:32

only my current laptop, but the previous laptop

6:35

and the laptop before that because all those backups

6:37

are still sat on the server. They're all

6:39

in one big folder, which

6:42

means it's super time consuming, copying

6:44

a whole bunch of files that are never going to change

6:46

because those laptops don't exist

6:47

anymore. So I think there's

6:49

ways I could improve this, but I'm

6:52

too far down the line of using it. I'd

6:54

have to just replace it completely, I

6:57

think. Right. And you said you

6:59

do your incrementals at like every six

7:01

hours. Do you have any idea

7:04

how long it takes your incremental backup

7:06

to run?

7:07

That's a good question. So the

7:10

most recent one that started at four o'clock today

7:13

took an hour to

7:16

RM the oldest alpha

7:18

backup, and then it took less

7:21

than a minute to move all of the

7:24

alpha backups up one. It moves, not copies.

7:26

Yeah. And then to copy alpha 0

7:28

to alpha 1

7:30

took an hour and 10 minutes.

7:32

And then actually doing the backups took 15

7:36

minutes, not long at all. So the actual

7:38

backup, the diff of what has changed

7:41

since midday to four

7:44

o'clock on all of those systems is 15

7:47

minutes worth of copying files over the network.

7:49

But the prep took over

7:52

an hour to do. So

7:54

it's horribly inefficient. Now bear in mind, this is running

7:56

on an eighth gen HP micro

7:58

server with four.

7:59

aging 4 terabyte

8:02

drives in them. I could probably

8:04

buy a Synology with one 12 terabyte

8:06

drive and it would just be like insanely fast.

8:09

But that's where we are. That's the system I've got.

8:11

And I don't really want to go buying a whole

8:13

load of more infrastructure for backups.

8:16

And I should note,

8:17

I probably didn't explain it particularly well. I

8:20

have got a blog post. If you just do a search

8:22

for Popi our snapshot, you'll

8:24

find my blog from a couple of years ago. Why

8:26

I explained this in much more detail about how it all

8:28

works and how it fits together. And I'll link it in the

8:30

show notes.

8:31

I too have used our snapshot very

8:34

happily for many years, but I'm here

8:36

to tell you, resistance is

8:38

futile. Borg Backup is here

8:40

to replace our snapshot. Okay. So

8:43

I have recently done the very

8:46

thing that you were posturing. I

8:48

have completely replaced my backup

8:51

solution with Borg Backup.

8:54

And I suppose it's important

8:56

to summarize why I decided

8:58

to replace our snapshot after so many

9:00

years of being happy with it. Is it anything to

9:02

do with the reasons that I gave about it being slow

9:04

and a bit of a conism? Yeah,

9:07

I thought it might be. So I

9:09

use our snapshot to do two types of

9:11

backups. One to do

9:13

backups of my home directory.

9:16

And that is to protect me from

9:19

when I do a silly and I go and delete

9:21

a bunch of stuff accidentally that I really didn't

9:23

mean to delete

9:24

and I need to recover it. And I

9:26

also use it to back up my servers, which

9:28

is mostly static

9:30

content. It's videos and music

9:32

and things of that nature and some other

9:35

stuff. And I'll get into that in a bit later.

9:37

But those workstation backups of my

9:39

home directory, I've got terabytes

9:41

of data and millions of files.

9:44

And our snapshot takes a long

9:46

time to run even on

9:49

very, very fast, all NVMe

9:52

storage solutions for the

9:55

source where the data resides and

9:57

the target to where it is being backed up.

9:59

So spinning rust is not

10:02

the only limiting factor

10:04

here. It is just a inefficient

10:07

process. So I wanted

10:10

to be able to take snapshots of

10:12

my data at much more frequent intervals

10:14

to give me more tight restore

10:18

points. Jason Valeeva More factor, silly factor.

10:21

For when I inevitably RM something

10:24

in the wrong place and destroy hours

10:27

of work.

10:28

So I was thinking I would

10:30

switch my file systems to ZFS

10:32

and get into ZFS replication. And

10:35

you know what? Life's too short for

10:37

that, you know, for

10:40

your laptops and your trivial stuff

10:42

at home, getting into balancing

10:44

ZFS and doing storage architectures

10:47

and all of this sort of thing. And I don't trust BtreeFS.

10:51

Send your hate mail to Joe at late night

10:53

Linux dot com. It's bitten

10:55

me and I'm not going back there and XFS

10:58

I love, but XFS

11:00

is missing some features that would be

11:02

good for a backup solution like there's

11:04

no native compression

11:07

on the file system, for example, something

11:09

that you can do with BtreeFS.

11:12

So I wanted to find a solution that enabled

11:14

me to keep XFS, which I

11:17

trust and I like, but

11:19

also faster than our

11:22

snapshot, more space efficient

11:24

than our snapshot. But you talked

11:26

about like how convenient the restore

11:29

is from our snapshot, because you can just go to

11:31

an any point in time version

11:34

and all of your stuff is there. Yeah. And

11:36

I love that. And I also love, as you say,

11:39

when our snapshot is configured, it

11:41

just comes along and it does its job.

11:43

And I wanted that. So bulk backup is all

11:45

of these things. So it is

11:49

very fast, much faster

11:51

than my snapshots. And I'll get to that in

11:53

just a moment. But the space efficiency

11:55

is interesting because it does the duplication

11:58

as well as.

11:59

compression. So when you create

12:02

a repository for your

12:04

backups, you can say what compression

12:07

you want to use on that repository.

12:10

So I've created two repositories,

12:12

one which is a target for stuff that is

12:15

probably going to compress well, like

12:18

my documents and general data.

12:20

And I have another one where I have no compression

12:23

applied, which is where the music

12:25

backups and the video backups and the

12:27

photo backups go to because you

12:29

get

12:29

little to no benefit from

12:32

recompressing those things. And it just adds

12:34

to the time the backups take.

12:37

In addition,

12:38

it has a very simple means

12:41

of encrypting your data.

12:43

And

12:44

at some point in the future, I want to get

12:46

to sending another

12:48

copy of my backups to some offsite

12:51

place. I've currently got the office in

12:53

town and the server here. So I have two

12:56

places, but I'd like something that's completely

12:58

separate as well.

12:59

And the restore process for Borg

13:02

is very elegant. You have

13:04

they call them archives, but

13:06

it's basically a point in time backup.

13:09

And you can just say, go and mount

13:11

that and it uses fuse mounting and

13:14

it suddenly just appears just

13:16

like a directory anywhere on your

13:18

system. And you can then use our sync

13:21

to copy a bunch of files

13:23

or the whole thing. But you don't,

13:25

it's not like the old days of

13:28

backup software, where you have some

13:30

way of saying, right, extract this whole

13:32

backup. You

13:36

just basically mount the

13:38

point in time backup that you're interested

13:41

in. And then you can restore some or any

13:43

files that you want out of there back

13:45

over the target, just like I would do

13:48

with our snapshot. So I love that.

13:50

So because of all of these reasons, it

13:53

has replaced our snapshot

13:55

for me,

13:56

and I'm using it for my workstation

13:58

backups to.

13:59

do regular snapshots of my home

14:02

directory. And

14:04

that also includes backing up my

14:07

Keybase folder. So I

14:09

have got data in Keybase

14:12

for projects that I work on with other

14:14

people where we keep secrets

14:16

that we want to share with members of the project.

14:19

But I also want to have backups of that stuff

14:21

in case Keybase ever disappears, right? So

14:23

it's able to peek inside there. But

14:26

the backup is encrypted. So, you

14:28

know, I now have

14:29

a great way to back that stuff up.

14:32

And my workstation backups are

14:35

many terabytes and many millions

14:37

of files. And my incremental

14:39

backups now take 30 seconds.

14:43

So I am now doing hourly backups.

14:46

So Mark was asking about retention

14:49

earlier, my retention for my home

14:51

directory is I have 24 hourly

14:54

backups,

14:55

I have seven dailies,

14:58

four weeklies, six

15:01

monthlies, and one annual.

15:03

Now I've only been using this for about four

15:05

weeks at the moment, so I don't have that history

15:07

of stuff yet.

15:10

And then I'm using it on the server as well. And

15:12

the server is doing, you know, backups of the music

15:14

and the videos, but what I also have is a headless

15:16

version of Dropbox running on the server.

15:19

And I backup our Dropbox

15:22

folder as well. So if

15:24

Dropbox ever goes away or something disastrous

15:26

happens, I now have our Dropbox

15:28

stuff backed up. And I think

15:31

what's worth pointing out here is that you

15:33

can completely automate Borg backup

15:36

itself from the command line in much the

15:38

same way you would with our snapshot.

15:41

But there are some front ends that

15:44

make it even more palatable

15:46

and straightforward to use. So

15:48

on my workstations, I'm using a bit of software

15:51

called vorta, which is

15:53

a graphical application that

15:56

is

15:57

dead simple to use and you up

16:00

archives very very quickly and

16:02

say what you want to back up. It

16:04

has a simple mechanism of excluding so I

16:07

touch a file in any folders that I

16:09

don't want to back up. I just create a hidden

16:11

file called dot no back up

16:13

and Borg sees that and goes ah well I don't

16:16

back this one up so if I've got things I want to

16:18

exclude it's easy to do. So I'm

16:20

using Vorta on the workstation.

16:22

It sits in the indicator area with all

16:24

of the other things and it

16:27

glows red when it's doing its backups

16:29

and you can click on it and if

16:31

you want to restore you

16:33

can just

16:34

click the in the list of backups

16:36

and say mount it and it just appears

16:39

as a fuse file system

16:41

in your file browser and you can go and poke

16:43

around and look in it and all that sort of thing.

16:45

That is super neat. I mean that Vorta

16:48

does look I mean the video on their home page

16:50

is on a Mac so I assume the Linux version looks

16:52

exactly the same. Right.

16:55

And as you point out this will work

16:57

on Mac, Linux and Windows.

17:00

So like you I only care

17:02

about Linux but if there are people out

17:04

there who have got a mixture of machines

17:06

in their home and they want a solution that

17:08

can work everywhere then Borg

17:11

plus Vorta

17:12

will give you that

17:14

and then on the server I'm using

17:16

a utility called Borgmatic

17:19

which basically presents as a

17:21

YAML file where you configure

17:23

what you want to back up and it's just a

17:25

much simpler interface

17:28

to drive Borg itself. So

17:30

that's what I'm using and I don't

17:33

have ten years of experience with it like I

17:35

do with R snapshot but so far

17:38

it's giving me the R snapshot

17:41

experience that I've enjoyed

17:43

but it's quicker faster using

17:46

less disk space as well. Nice.

17:49

And when you have Borg

17:51

installed on remote endpoints

17:54

over SSH it can

17:56

accelerate the backups over

17:58

SSH by virtue of the

17:59

of having Borg on both ends

18:02

of the connection. So it's not

18:04

just doing like a file system

18:06

transaction. It actually has a protocol

18:09

between the two, the client and the server.

18:11

So it's very, very fast remotely

18:14

as well. Is

18:15

Borg installable as a like

18:17

Deb on Linux and as a packages

18:19

for everything? I'm going to say, yes, I'm obviously

18:22

using it on NixOS.

18:23

It's working fine here. I have seen

18:25

that the guides talk about the many Linuxes

18:28

that it's available for. Cool.

18:30

And it's sponsored by

18:32

Borgbase.com who are basically

18:35

a hosting company that offer you remote

18:38

storage and also Hertzner

18:41

offer storage box solutions, which

18:43

are Borg

18:45

enabled. So if you want an

18:47

offsite solution, there are hosting

18:49

providers that support the project

18:52

by providing you with cloud-based

18:54

storage that is accelerated for Borg. Okay.

18:57

Sold.

19:02

Linux matters is part of the late night Linux

19:04

family. If you enjoy the show, please

19:06

consider supporting us and the rest of the late

19:08

night Linux team using the PayPal or

19:10

Patreon links at linuxmatters.sh slash

19:13

support

19:14

for $5 a month on Patreon. You can

19:16

enjoy an ad free feed of our show or

19:19

for $10, get access to all the late

19:21

night Linux shows ad free.

19:23

You can get in touch with us via email show

19:26

at linuxmatters.sh or chat

19:28

with other listeners in our telegram group.

19:30

All the details are at linuxmatters.sh

19:33

slash contact.

19:37

My backup solutions a bit different

19:39

because I think the main use case I'm thinking

19:41

of is a bit different. I'm not thinking

19:44

about backing up entire machines

19:46

or my entire home directory all the time

19:49

in case I've messed something up. I'm more thinking about

19:51

what are the things that are crucial

19:54

for my family to make

19:56

sure that we never lose, which are things like

19:58

our family photos. And currently

20:01

we have quite a good system within the house

20:03

whereby our photos are

20:05

instantly uploaded from our phones to our next

20:08

cloud server and synced between all

20:10

the devices which talk to that. So we've got plenty of copies

20:12

of them around the house. If one machine goes pop, nothing's

20:14

getting lost. But

20:16

were there to be a complete disaster and all the machines get

20:18

lost somehow, I wanted to have some sort

20:20

of further offsite backup that

20:22

I could do. And I've been thinking about the same

20:25

sort of things that we've already been talking about,

20:27

not just how do I back it up and do it easily

20:29

and make sure it's encrypted, but also what's

20:32

the recovery process like? How

20:34

easy is it to test the recovery process after

20:36

just doing your first backup?

20:38

And I recently discovered a tool called

20:41

R-Clone. And R-Clone is

20:43

a similar sort of tool. It will

20:45

synchronize files from machine to

20:47

a machine. But the way it does it is quite

20:49

clever and really flexible and gives you a lot

20:52

of interesting options to cover a lot of different

20:54

cases.

20:55

So the way it works is it gives you

20:57

a load of essentially options

20:59

for pluggable backends. And

21:01

that could be somewhere on a file system.

21:04

It could be somewhere over SSH

21:06

or SFTP or any other

21:08

sort of network protocol. But it could

21:10

also be

21:11

any other way there is of storing

21:14

files remotely. And that could include

21:17

something like S3 object storage,

21:20

even something like WebDAV or

21:22

IMAP or anything which could

21:24

feasibly store a file, you could create

21:26

an R-Clone backend to do it.

21:29

So in my case, I've gone with an

21:31

S3 compatible option. I've actually gone

21:33

with iDrive E2 storage because

21:36

they sponsor R-Clone. And I

21:38

thought if I'm going to give money to someone, I may as well

21:40

give it to people who support the project. So

21:42

that's good. Yeah, you can just use the

21:45

generic S3 backend and point

21:47

at any provider that supports that API

21:49

or there's a few which support a version of the

21:51

API like backblaze that have their own

21:54

backend. But because this is all private

21:57

family stuff, I wanted to make sure it was encrypted. And the way it works is

21:59

it's a very simple process. way that RClone handles

22:01

things like that is as well as these things which are actually

22:04

a backend for storing files somewhere, there's

22:07

a set of sort of meta backends for

22:09

doing other things. So there's one called

22:11

Union, which will take a load of backends

22:14

and stick them together as though they were one backend.

22:16

So you might have a load of small accounts

22:19

across various providers, but

22:21

you want to just back up to them somehow,

22:24

use that storage as a blob, or

22:27

have multiple accounts and back

22:30

up your stuff to all of those at once.

22:32

And there's your backends for doing that by

22:34

putting them in front of the real backends. And

22:37

similarly with encryption, the way that you do that

22:39

is you create your backend for where you're actually

22:41

storing the files, then you create an encryption backend,

22:44

which points to the real backend, and then

22:46

you back up to the encryption backend, which

22:48

then transparently does all of

22:50

that basically. And the process

22:52

is exactly the same. Like you've got all of these commands

22:55

in RClone for doing

22:56

various things and they all work the same, whichever

22:59

backend you're using. Nice. I've

23:01

been aware of RClone for some time. And

23:03

it's been on my list of things to take

23:06

a proper look at and I've never got round to it.

23:08

But the way that I've understood it is

23:10

it's like Rsync for

23:12

cloud storage. Yes. I mean,

23:15

that's basically how it operates, but it's

23:18

cloud storage and any other storage. You

23:21

could equally use it with another server

23:23

on your LAN or just files

23:25

locally if you wanted

23:26

to. So you could use it like Rsync?

23:29

Like Rsync, yeah, exactly. So the key

23:31

of the question is what happens if you need

23:34

to restore a file? How do you get stuff

23:36

back? Right. So the backing up is simple.

23:38

I've got a cron job which says, take

23:41

this directory and clone it to this

23:43

backend. And

23:45

that's all I need to tell RClone. It does that however

23:47

often I want it to. And you see, if

23:49

you look at the bucket in the object storage,

23:51

you just see a bunch of encrypted file names.

23:54

But here's the clever bit. Like

23:56

Borg, you have the option on your

23:59

machine where RClone is running to

24:01

say mount this back

24:04

end and you get it mounted as a directory.

24:07

So you can do that. You can also

24:09

say take this back end and expose it

24:11

over whatever remote protocol.

24:14

So you could have it on your server. You could have RClone

24:16

running on your server, but say, actually give me that

24:18

as a web dev mount.

24:20

And then from somewhere else on your network, you could

24:22

do the same thing. You could also take

24:25

the same config file from the server

24:27

that's doing the backups and put that

24:29

on another machine like your laptop

24:32

and then run the same command to mount that

24:35

back end. And it'll do that on

24:37

your local machine.

24:38

But aside from doing all of this on the command

24:40

line, you also get the option of running

24:43

a web GUI, which is actually what I used to

24:45

do the configuration.

24:46

So you have an interactive CLI for

24:49

adding rebotes and then the CLI commands for

24:51

running it, but you can also do it all through quite

24:53

a decent, although it describes itself as an experimental

24:56

web front end, which you run.

24:58

You

24:58

do RClone, RCD, dash dash

25:01

web GUI on whichever machine has

25:03

the config file that you

25:05

want.

25:06

And then as well as setting it all

25:08

up like that, you get the option to

25:11

browse any of the remotes.

25:14

So I can go in there and I can say show me my

25:16

backup destination and it will show me

25:18

the real file names and the real files. And then I can

25:20

just click on one and download it, or

25:23

I can basically manage all of the mounts

25:25

that I've got. So I can from there, I

25:27

can say mount this and then that appears in my file

25:29

manager and I can browse it.

25:32

At the moment, in terms of historical backups,

25:34

the way I've approached that is using

25:36

S3 versioning. So if I have

25:39

files that I delete, they'll

25:41

be retained as a version or if they get updated,

25:44

I'll get a new version of them in the bucket,

25:46

which I can then expose

25:48

through RClone, the old versions as

25:50

well as the current version.

25:52

This is still evolving. I've only

25:54

put this in place recently and I'm currently only doing, you

25:56

know, just our family photos

25:59

at the moment to see how it goes. and how much it's going to cost.

26:01

But it's certainly something I'm considering expanding

26:03

to the other important things.

26:06

I find it interesting that you don't care so

26:08

much about non-family

26:11

photos or your home directory

26:13

or anything like that. If my home

26:15

directory blows up, I can reinstall and

26:17

start again. I don't handcraft

26:20

my dot files like Martin does.

26:23

I used to do that, but now I don't have to because

26:25

I have everything powered by Nix. So

26:29

when I come to restore a recovery

26:31

machine, which I did recently, I reinstalled my laptop.

26:34

The install process gives

26:36

me an install fully configured, including

26:40

all of the directories in my home

26:42

directory where my data lives. So

26:44

then I just imported my Vorta backup

26:47

profile and ran a restore

26:49

and I was back up and running. And that whole process

26:52

from install to fully restored system

26:54

was under an hour. Which is

26:57

really nice. But like you, Mark, the

26:59

server backup is backing up our

27:01

important stuff. So we have sync thing on all

27:03

of our laptops with our documents

27:06

and our photos. So like you, everything's

27:08

in multiple places. But

27:11

because we've now got this magical internet LAN,

27:14

if I delete something in sync

27:16

thing, it almost instantaneously

27:18

gets deleted from everyone

27:20

else's machine, which is why I wanted

27:22

this bulletproof backup sitting

27:24

on top of it.

27:25

It's interesting that you say about installing

27:28

your system from scratch and then restoring

27:31

from the most recent backup that you have in bulk. And

27:33

now I think about it, if I get

27:36

a new laptop,

27:37

I do a clean install and I cherry

27:40

pick a few things out of the

27:42

backups of my previous laptop, like my music

27:44

collection, for example, and

27:47

maybe a couple of other folders and everything else I sync

27:49

thing for. And I'm thinking, I actually

27:51

back up a tremendous amount of stuff that I am never

27:53

ever going to restore. Like

27:56

there's a whole truckload of stuff,

27:58

like the dot cache folder and

27:59

I am

28:02

almost guaranteed never to want

28:05

to restore an old browser

28:08

profile cache folder. That's

28:10

just not going to happen. So I

28:12

think I need to optimize my backups a little

28:14

bit better. So I used to do exactly

28:17

what you're describing with our snapshot. I would just

28:19

say back up my home directory dot files,

28:21

the whole thing. And then, as you

28:24

say, be selective in the restore process.

28:26

I only need the documents folder in the downloads

28:28

folder and this and that.

28:29

But now with Borg, I've

28:32

changed my strategy to say these

28:34

are the directories that have data in it that I

28:36

want to back up and it really is data, you know,

28:38

is user data not configuration

28:41

anymore, definitely not caches. So

28:43

when I do a restore, I know it's that's all

28:45

of the stuff that I want. I don't trust myself.

28:47

I don't trust myself that I would

28:50

pick all the right folders and then somewhere down the line,

28:52

I would create a folder in my home and it's

28:54

never been backed up. That's my problem. I

28:56

don't trust myself to put things

28:58

in the right place.

28:59

Maybe I need more discipline with my backups.

29:02

So that's how we're all doing our backups.

29:05

Is anyone else out there doing it differently? We'd

29:07

be interested to hear what your solutions are.

29:09

In particular, I would like to hear what

29:12

people are doing for backing up

29:14

and viewing their family photos. I

29:16

have a suboptimal solution at the moment. Your

29:19

help finding a new solution would be most

29:21

welcome.