129 lines
10 KiB
Plaintext
129 lines
10 KiB
Plaintext
|
|
Episode: 30
|
||
|
|
Title: HPR0030: Network Backups
|
||
|
|
Source: https://hub.hackerpublicradio.org/ccdn.php?filename=/eps/hpr0030/hpr0030.mp3
|
||
|
|
Transcribed: 2025-10-07 10:28:09
|
||
|
|
|
||
|
|
---
|
||
|
|
|
||
|
|
.
|
||
|
|
Hello and welcome to another edition of Hacker Public Radio.
|
||
|
|
Today I'm your host, D.J. Dawesman, and today we're going to be talking about backups.
|
||
|
|
I'm going to kind of dive into a couple of different types of backups.
|
||
|
|
I'm going to cover a couple of different pieces of software for one type of backup.
|
||
|
|
I'm going to cover and talk about the hardware setup and then kind of go over some other
|
||
|
|
concepts and such.
|
||
|
|
First, the backups that I used to make early on of my systems were basically you power
|
||
|
|
the system down, you boot it off in floppy disk with Norton Ghost or Clonezilla or something,
|
||
|
|
and then you either burn it to a local CD or DVD or you can shoot it over the network
|
||
|
|
to your ghost cast server or whatever you're going to do with it.
|
||
|
|
That's nice.
|
||
|
|
That gets a full complete backup of your system.
|
||
|
|
However, the downside to that is it's pretty manual and your system has to be down for
|
||
|
|
this to happen.
|
||
|
|
Because of that, I don't think a lot of people do those often enough.
|
||
|
|
I know I don't.
|
||
|
|
What I really like and what I've come to really appreciate about what we use at work is
|
||
|
|
an online backup system that backs up stuff over the network and automatically tracks
|
||
|
|
incrementals and runs while the system's up, so your production system so I have to
|
||
|
|
be down.
|
||
|
|
I started looking around for an open source way to do this on my own systems at home.
|
||
|
|
What I came across were two different packages, one is called Amanda and that's been the
|
||
|
|
good old standby for, I don't know, 15, 20 years now, and basically it's called Amanda
|
||
|
|
for originally created by the University of Maryland, I believe, and basically it's
|
||
|
|
a loosening collection of different scripts and programs, all packaged together with
|
||
|
|
a server and basically it runs as a Damon and so when it runs, just every night, it
|
||
|
|
just gains your system for files that have changed and that's your incremental backup
|
||
|
|
because it shoots off files that have changed on each of your servers over to a central
|
||
|
|
storage backup system.
|
||
|
|
Of course, the server component on the backup system then saves automatically manages writing
|
||
|
|
all the backups to tape for you, typically with a larger installation you're going to have
|
||
|
|
a array of multiple tape drives and a tape library, a robot moving tapes in and out.
|
||
|
|
You can also do that, they can store, I don't know if Amanda does this yet, stores to local
|
||
|
|
storage pools on online hard drives basically, so you don't have to have the tape drives for
|
||
|
|
the backups, but Amanda has been the standby for people doing stuff like that and it mostly
|
||
|
|
is intended for working with Unix or Linux systems, Windows systems, I think there's kind
|
||
|
|
of a hack way to do that using SMB shares and so on, to get backups of your Windows systems
|
||
|
|
using Amanda, I really wanted something a little less clumsy and I finally stumbled
|
||
|
|
across another package called Backula and their slogan is, it comes by night and sucks the
|
||
|
|
vital essence from your computers and it's only, it came about in 2000, 2001 or so I think
|
||
|
|
and it's, all of a sudden it has some very strong development behind it and a surpassed
|
||
|
|
Amanda quite a bit by its features, capabilities and robustness I think, which is kind of
|
||
|
|
unfortunate for the Amanda folks, it's a good, Amanda's a good thing, but anyway this
|
||
|
|
is Backula product, it's basically all completely distributed so you can have like a single
|
||
|
|
tape drive on each, you know, of several different computers and you can have those act as
|
||
|
|
agents that receive, you know, you have a central backup server that kind of, you know,
|
||
|
|
pulls the strings and manages all the backups to these other systems that have the storage,
|
||
|
|
all your client systems back up to the main Backula server and you could also just have
|
||
|
|
a standalone Backula system if you have multiple tape drives hanging off this one system
|
||
|
|
or a tape library, so that's what it does.
|
||
|
|
Commercial products in the same area of course if you're familiar with TSM or Legato
|
||
|
|
network or ArcServe backup exec, these are all very, very expensive commercial products
|
||
|
|
that basically do what I just described network backups with your systems online, so anyway
|
||
|
|
I decided I wanted to do something like this on my own for my own systems because after
|
||
|
|
a while you get so many different laptops, computers running and stuff, you don't keep
|
||
|
|
good backups of them and so I also just happen to have an XABXB210 dual drive 10 cartridge
|
||
|
|
didn't tape auto-changer that I acquired and I still got to get some tape drives to put
|
||
|
|
into it but basically any 8 millimeter tape drive will work with this thing that will
|
||
|
|
be compatible with the robot, the gripper, it's pretty slick, I finally, I've had it for
|
||
|
|
a couple of years but when I decided I never actually hooked it up and actually I like
|
||
|
|
used it and so I sat down and basically to get an auto-changer to work with Linux, there's
|
||
|
|
a couple things you're going to have to do, obviously you've got to get your scuzzy card set up,
|
||
|
|
I was using a future domain PCI card which worked fine under Fedora Linux, Fedora 5 and it's
|
||
|
|
pretty old but I've got a system that I'd use that's Fedora 5 and for experimentation and stuff
|
||
|
|
and at work fine I could control the accessor but under sent OS 5 which is basically Red Hat
|
||
|
|
Enterprise Linux, I have not been able to get the thing to work and the driver keeps crashing
|
||
|
|
anytime I try to use it, I don't know if it's an application problem or driver problem on
|
||
|
|
kind of suspecting it's a driver problem at this point because sent OS didn't come with this
|
||
|
|
driver by default whereas Fedora did, I was able to get it to compile but anyway some problems I've
|
||
|
|
had, so basically how do you use an auto-changer with Linux? Of course you mod probe your scuzzy card,
|
||
|
|
most tape libraries use the scuzzy generic Linux device driver set and so that rides on top of
|
||
|
|
your above your scuzzy card stack and then also you're going to have to have a scuzzy tape or ST
|
||
|
|
driver for your tape drive and your scuzzy changer is going to be a scuzzy idea on your scuzzy bus,
|
||
|
|
your tape drives will be separate IDs also on the same bus so basically you can
|
||
|
|
cat slash proc slash scuzzy slash scuzzy and if your scuzzy drivers initialize properly you'll see
|
||
|
|
both the auto-changer and the scuzzy tape drive and from that point you're good to go to try
|
||
|
|
with the MTX command and you can check out the man page on that for flags and how to use it
|
||
|
|
obviously and you'll reference the scuzzy generic device. Once the SG module is initialized into
|
||
|
|
the kernel you'll have a couple of devices out in your dev file system and usually like you'll
|
||
|
|
have at least SG zero and SG one those will represent the auto-changer and the scuzzy tape even
|
||
|
|
though you won't access the tape with the scuzzy generic interface to it. You'll use your regular RRMT
|
||
|
|
or whatever little device you have that your OS provides for that. The thing about scuzzy generic
|
||
|
|
is basically just very low level scuzzy interface so it just lets commands throw data at the scuzzy
|
||
|
|
bus, throw commands and stuff and so that's how that's why auto-changers are typically accessed
|
||
|
|
with the scuzzy generic. So once you get those things working then the next layer the actually user
|
||
|
|
commands is going to be MTX and there's a couple different user space commands for accessing
|
||
|
|
your tape-changer but MTX is pretty much the good old standby that I think everyone uses Amanda
|
||
|
|
uses MTX in order to manipulate the auto-changer or backulate uses MTX and I think they can use some
|
||
|
|
other utilities too but MTX that's that's the good way to go and it'll do some verification you
|
||
|
|
can do some basic status information from your library. I've got a little script when I had it
|
||
|
|
working under Fedora I could have it just you know run the move the robot back and forth and move
|
||
|
|
tapes automatically just kind of an exercise so that was a lot of fun to see working. So like I
|
||
|
|
said you were talking describe some of the applications software that writes on top of it so so
|
||
|
|
anyway once you get these layers of things set up and you get your your applications software
|
||
|
|
up either Amanda or backulate or possibly something else then basically you should be able to have
|
||
|
|
nightly incremental backups of all your files. I for most part I you know I've got a lot of like
|
||
|
|
text notes that I have personal documentation that I write for myself. I've had a growing library
|
||
|
|
of like video and photo photos and stuff that need to be archived a little bit better than I
|
||
|
|
currently do. So that's that's some of the things that I intend to use with it. I currently
|
||
|
|
I'm having some problem. I haven't actually gotten backulate up and running yet but that's mostly
|
||
|
|
because I once I I switched mid-mid process to a new system I got a dedicated piece of hardware
|
||
|
|
just to run backulate on and unfortunately like I said I've been having some problems with sent
|
||
|
|
who has but once I can get those issues resolved and I'll go ahead and start working again with
|
||
|
|
backula and the thing with with Amanda and backula and the I mean these the trade-off of having
|
||
|
|
a very convenient nice backup server is it takes quite a bit of configuration and massaging
|
||
|
|
to get these things to work right. You've got you know exclude lists like there might be certain
|
||
|
|
directories on your system that you don't want backed up. I don't know things like obviously
|
||
|
|
you don't want it to mess with proc there's no reason for it to be touching anything under there
|
||
|
|
and other other file systems and such. So it's going to take some tuning there's going to be quite
|
||
|
|
a few concepts you're going to have to I get your your head wrapped around and backula try is
|
||
|
|
pretty hard to explain these concepts early on to help you make the right decisions to configure
|
||
|
|
the system what you need. Some of the Amanda documentation really seems to be kind of they're
|
||
|
|
switching from their older style of documentation to a wiki and it seems like there's a lot of stuff
|
||
|
|
it is just really in the middle of a transition right now. My apologies to the Amanda folks if
|
||
|
|
they've worked out all those issues since I've looked at it but it seems like they're unfortunately
|
||
|
|
they were kind of overtaken it seems like with the functionality and backula there but it so that's
|
||
|
|
basically a good robust backup system how you would use it how to set it up. I'll have some
|
||
|
|
notes on the hacker public radio for some of these different things I've talked about.
|
||
|
|
So anyway have a good day enjoy and I'll see you later.
|
||
|
|
Thank you for listening to hacker public radio. HPR is sponsored by caro.net so head on over to
|