Very interesting insight on these options!
1 Like
Interesting definition you invented there. The definition you provided a screenshot of said no such thing. You seem to be conflating ‘backup’ with ‘archive’… or something else. Hard to tell.
No, it doesn’t. Unless you are in the habit of running rsync --delete after every keystroke, that doesn’t (can’t) occur. Since no-one who is sane runs rsync --delete after every keystroke, your strawman has fallen over.
No-one is talking about Raid 1 here. We’re talking about daily backups. You have up to 24 hours to recover files that slipped past (weren’t covered by) version control. If important data can actually go missing on your system for 24 hours without you noticing and doing something about it, then PEBCAK.
Anyway… as I don’t think this ‘argument’ is a) on-topic, b) constructive, or c) heading anywhere useful, I think I’ll take my leave so as not to further derail the thread. Keep your data safe what ever way works for you, and have a nice day.
Anacron only runs once a day. Given that borg backups take less than a minute to create I like to make them more frequently.
I’ll have to take a look at the systemd units.
just dd commands
sometimes as .IMG (disk image files) - giving more control, other times entire cloning of drive
pretty much it
My /home is sync’ed via a Nextcloud server into a FreeNAS. This allows me to access my files and email (personal server) anywhere in the world by PC or phone.
The FreeNAS has 28 days of snapshots so I can use that to get a file back I have deleted. This is the on site backup.
Then at 03:00 it is sync’ed again into an AWS S3 bucket via script using AWS CLI. This is the off site backup.
This keeps additional versions for 7 days and then deletes them.
I just write routines I repeatedly use into shell scripts. I like using one script to execute several other scripts for routine maintenance (like Apple’s periodic daily/weekly/monthly scripts).
This also makes modification and adaptation to other distributions very simple. I am going to be pulling packages through the manager on each OS whenever I reinstall anyway, so I just start from scratch every few months or longer if I am busy.
From a clean install of Fedora for example, I have about 5 main scripts with a reboot between each one before I am in an optimized environment for what I do.
I use Veeam Backup & Replication for everything. Price is a bit prohibitive for private use, but for any sort of business use their pricing is very reasonable.
This is only in regards to linux. Please don’t clutter the discussion with other operating systems.
1 Like
Sorry, didnt see the Linux category, my mistake.
2 Likes
So (elsewhere) you’re saying you’re not talking about RAID1? So don’t call what you’re doing a “mirror” then; it’s a copy or a backup. By incorrectly using the word “mirror” you’re implying that you’re constantly running your rsync to produce a near real-time “copy” of a disk. Bear in mind that some of us have spent hours explaining in simple words to PHBs that a mirrored disk is not a backup.
No, but some methods are sub-optimal. And deleting old files from your backup within 24 hours is definitely sub-optimal. You are free to decide it’s reasonable for you, but it’s perfectly reasonable to point out there’s a significant problem with it (even if you don’t see it as a problem).
1 Like
No-one was ever talking about Raid 1. The phrase never got used. If you think anyone was talking about Raid 1 then you are simply projecting.
It is called a mirror, and I (along with the general population) will continue to call it a mirror, because a mirror is exactly what it is. Heck, even the rsync man page says it is used for mirroring.
A ‘mirror image’ is a perfect, identical copy of something. That’s the meaning it has always had. It’s also defined as a carbon copy, clone or duplicate. When you say “it’s a copy or a backup” you are right as well. It’s all of those things. Welcome to the wonderful world of synonyms where people are free to describe something with different words because they all have the same meaning in given contexts.
I’m not implying any such thing. In fact, given that I previously wrote “Since no-one who is sane runs rsync --delete after every keystroke” it should be pretty clear that’s not what I’m implying. Please project less, read more.
Well perhaps you should stop then because you are wrong and wasting your time.
You are trying (for whatever reason) to limit the historic and general meaning of the word ‘mirror’ to Raid 1. ‘Raid 1’ is neither a definition nor a synonym for ‘mirror’. Never was, isn’t, nor will it ever be. Raid 1 is merely one, specific example of mirroring — nothing more. Mirroring isn’t Raid 1. A ‘Raid 1 mirrored disk’ isn’t what this discussion has ever been about. This discussion is about rsync, not Raid 1. May I suggest you stop projecting your Raid 1 issues onto rsync — they are different things.
There has never been — and is no — time requirement for mirroring. ‘Near real-time’ is a condition that you have decided to personally place on the term for [waves hand] some arbitrary reason.
Like the other guy, you are trying to redefine a word to suit your own agenda — whatever that may be. You can’t do that. That’s not how ‘language’ works. You can’t redefine ‘dog’ to exclusively mean ‘golden retriever’ for exactly the same reason. That’s an English fail.
No matter how much you desperately want ‘mirror’ to exclusively have a more specific meaning — it doesn’t. “To mirror data” has always meant “to make an exact copy of data” — and it will always have that meaning. How frequently that mirror is refreshed is irrelevant — and not up to you to decide or define. Sorry.
Mirrors are a type of backup. Always have been. Always will be. Because ‘English’.
Once again, as I don’t think this ‘argument’ is a) on-topic, b) constructive, or c) heading anywhere useful, I think I’ll take my leave so as not to further derail the thread. Keep your data safe what ever way works for you, and have a nice day.
Whether a particular backup method is ‘optimal’ or not depends on the data that is being backed up and the circumstances surrounding it. You, personally, have no clue what my data is like, or how and when it changes, so you, personally, are in no position whatsoever to judge what ‘optimal’ is, nor what “sub-optimal” may be. Thus your definition of the word ‘optimal’ is meaningless, and that makes your definition of “sub-optimal” also meaningless — and your assertion completely invalid.
The same logic applies to every other system in the world. Unless you personally have intimate knowledge of it, you simply don’t have a clue what ‘optimal’ or ‘sub-optimal’ are. You’re just making assertions based on assumptions — and we all know what that means.
In my situation, deleting obsolete files and keeping my (failure-free) backups at a predictable size is exactly what I want. It’s ideal for my circumstances — the circumstances that you are ignorant of. If you are afraid of deleting files, and think that allowing old, obsolete files to gradually fill up and choke your backup drive is a good idea, then that’s your opinion. Good luck and have fun with that.
Besides, backup strategies do not need to be ‘optimal’ — they merely need to be ‘adequate’. Wasting time, effort and resources on overkill/complex backup strategies demonstrates a lack of wisdom — as far as I’m concerned.
Folks should develop a customised backup strategy that is adequate for their circumstances and data. I believe that this thread is (should be) about exploring different ways of doing that — not trying to push our personal agendas down other people’s throats.
2 Likes
I agree each persons back up needs are completely different from someone else.
for example my backups are daily and weekly
daily backup of active documents to internal secondary drive.(I never store my saved files on the os drive any longer than what is needed to complete them) and weekly image backups of system on the large external drive.
and os backup on a redundant external drive.
by doing it this way I am able to recover quickly from any situation.
in the event my os dies its simple to swap out the drive and be back up and running within a few minutes.
system image backups are only deleted if they exceed 4 copies(and then its the oldest ones that get deleted.
I know this is an old topic but i believe it should be re opened!
1 Like
This is an interesting thread. I looked at some of the recommended tools and noticed some details not mentioned:
- Timeshift specifically states it’s meant to be used like Windows System Restore for backing up system files, not personal files. Of course personal files could be included to be backed up, but that isn’t the intended use according to the github page:
Timeshift is similar to applications like rsnapshot, BackInTime and TimeVault but with different goals. It is designed to protect only system files and settings. User files such as documents, pictures and music are excluded. This ensures that your files remains unchanged when you restore your system to an earlier date. If you need a tool to backup your documents and files please take a look at the excellent BackInTime application which is more configurable and provides options for saving user files.
-
Borg’s developer mentions that due to its deduplication of data (great feature), it has no redundancy and is thus especially vulnerable to data loss if any part of the medium it is stored on suffers corruption or failure. Therefore it is recommended to have multiple copies, on another medium, in the cloud, etc.
-
Borg is available in Fedora’s repos (
borgbackupandborgmatic), but I did not find BorgWeb (a browser-based gui frontend). -
To my understanding, these solutions backup files, but not the filesystem structure, partitions, encryption, i.e. LVM, LUKS, etc. It seems for that there’s Relax and Recover (
rearin Fedora’s repos): http://relax-and-recover.org/
I’m still reading the documentation on Borg and Relax and Recover to see if they’ll fit my needs and how to set them up. It would be nice if there’s an easy to use, foolproof approach that combines the functionality of these tools.
1 Like
The --backup-dir option in rsync is handy for maintaining a mirror without losing data, although snapshots on the backup destination are arguably a better option.
For the past few months, I’ve been using rclone in place of rsync so that I can use the same script to run local, network (ssh) and cloud backups.
Some more info on the Borg universe:
Perhaps a basic question that doesn’t get much focus is what do you back up?
- Only system files (eg. System Restore)
- Only user files (eg. home folder, config files)
- Entire OS (eg. including system files and user files)
- Disk image (eg. including OS, partitions, bootloader)
0 voters
Most solutions discussed seem to focus on user files. Borg can backup the OS, but does not seem intended for that.
This means it takes some thought to decide what should be backed up, and requires some extra work to restore a system. On the other hand a disk image would be perhaps easiest to restore, but would need a lot more storage. And surely there are many other pros and cons to each approach.
2 Likes
Mostly just backup configs and scripts, anything else is mirrored across multiple devices in multiple places and I can just pull stuff in from those. Still waiting for a backup solution I trust that is a one time payment for a few TB of storage that is actually meant for backups.
Not like the other stuff I back up needs anything more elaborate than to be put on DVDs in a few places and archived in my google drive since it’s just the family pictures.
Isn’t a NAS that you put “somewhere else” exactly that? You buy it once, get as much storage as you need and are done. You can put it at your parents house, or a friends place, or at work. Get a DynDNS entry and you have a offsite backup that you can trust for a one-time cost. Including upgrade path.
Thanks! I just because of this thread started looking into Borg. This makes it super easy.