summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorbreadcat2024-08-30 15:35:58 +0100
committerbreadcat2024-08-30 15:35:58 +0100
commitc69c7fed461d268346ec4eba6157977382d93661 (patch)
tree67d74934086604b32d307bc63de990e11af068c5
parentdbf8a5c06149d444927ebff2f0f8a8ea13892266 (diff)
downloadblog.minskio.co.uk-c69c7fed461d268346ec4eba6157977382d93661.tar.gz
blog.minskio.co.uk-c69c7fed461d268346ec4eba6157977382d93661.tar.bz2
blog.minskio.co.uk-c69c7fed461d268346ec4eba6157977382d93661.zip
Additional post
-rw-r--r--content/posts/my-rudimentary-backup-process.md71
1 files changed, 71 insertions, 0 deletions
diff --git a/content/posts/my-rudimentary-backup-process.md b/content/posts/my-rudimentary-backup-process.md
new file mode 100644
index 0000000..99f15cf
--- /dev/null
+++ b/content/posts/my-rudimentary-backup-process.md
@@ -0,0 +1,71 @@
+---
+title: "My rudimentary backup process"
+date: 2024-07-26T09:56:00
+tags: ["Guides", "Infrastructure", "Linux", "Servers"]
+---
+
+I've heard of the 3-2-1 backup process, however I can't easily, securely (or cheaply) keep an offsite backup so I opt for more of a 1-2 process. I keep an offline store that's manually synced up every month or so.
+
+# The problem
+
+ As you can imagine though, there are a few problems with this... the moving parts involved are:
+
+* [NAS](/hardware/) - Contains my live data, but can only accept 1 hard drive (hardware fault), can't be hot-swapped, doesn't have a screen or keyboard connected and only has USB2 speed ports for data transfer.
+* [Desktop](/hardware/) - Connected via gigabit LAN, and has spare SATA ports for connection, however runs Windows.
+* [Laptop](/hardware/) - Ancient Thinkpad with USB3, running Linux however connects to the network using Wi-Fi and USB-SATA adapters are always a bit unreliable.
+* Backup hard drive - An offline 12TB SATA drive with the current backup.
+
+# The Solution
+
+So, with the above outlined, I feel as though the solution I came up with was quite elegant. We're going to use rsync on the NAS to send changes to the backup drive connected to our main Workstation computer (temporarily) running Linux from a live environment. Phew.
+
+## Destination
+
+First [boot the Arch Linux ISO](/whats-on-my-usb-key/) on your destination machine with the destination disk attached to it. I have to hammer `F11` on my machine but it varies.
+
+Once fully booted, check the environment you're in, mount your drive and set up a password for SSH access:
+
+```
+lsblk --fs
+mount /dev/sdb1 /mnt
+passwd
+whoami
+ip a
+```
+
+With the above details confirmed, it's time to begin the real process.
+
+## Sending
+
+Using the laptop, ensure you're able to connect to both the NAS, and then from the NAS, connect the desktop.
+```
+ssh nas
+ssh root@192.168.1.4
+```
+
+If you've already run the process previously, SSH verification will fail and you'll need to remove the keys from your `.ssh/known_hosts` file either manually, or via `ssh-keygen -R 192.168.1.4`.
+
+Now we can send our files, on the NAS, run the following (inside a `tmux` session if you think it's going to take a while):
+
+```
+rsync -avhP --delete --exclude=lost+found/ /mnt/ root@192.168.1.4:/mnt
+```
+
+Once complete, you've got an up to date backup and can shut down your live environment and disconnect the hard drive for another time.
+
+# Error handling
+
+Once, after a _very_ unscheduled power outage, my next backup failed with a data integrity error. I assumed this was the destination drive throwing the error as I'd had no errors reported on the source, however this appeared to be a silent failure. Only one file was affected and I was able to redownload the affected file without too much hassle. My process went a little something like...
+
+Stopping the NFS server and unmount the volume:
+```
+sudo systemctl stop nfs-server.service
+sudo umount /mnt
+```
+
+Using `fsck` on the volume, fixing any errors it finds, then re-mounting and restarting our NFS server:
+```
+sudo fsck.ext4 /dev/sdb1 -fy
+sudo mount -a
+sudo systemctl start nfs-server.service
+```