From 0e9fb8af5edd309ce65fc3904b449339a3c0abfb Mon Sep 17 00:00:00 2001
From: RAGHU VAMSI DORADLA <raghuvamsi.doradla@cumulocity.com>
Date: Wed, 19 Nov 2025 17:36:03 +0530
Subject: [PATCH] Disaster recovery for k3s tool based installations

---
 .../backup-and-recovery.md                    | 95 +++++++++++++++++++
 1 file changed, 95 insertions(+)
 create mode 100644 content/edge-kubernetes/edge-operations-bundle/backup-and-recovery.md

diff --git a/content/edge-kubernetes/edge-operations-bundle/backup-and-recovery.md b/content/edge-kubernetes/edge-operations-bundle/backup-and-recovery.md
new file mode 100644
index 0000000000..589c9a616b
--- /dev/null
+++ b/content/edge-kubernetes/edge-operations-bundle/backup-and-recovery.md
@@ -0,0 +1,95 @@
+---
+weight: 10
+title: Backup and Recovery
+layout: redirect
+---
+
+# Cumulocity Edge on K3s — Backup and Restore Guide
+
+This runbook describes how to capture and restore a **Cumulocity Edge** deployment that runs on **K3s** (installed via the `c8yedge` tool). Follow the numbered steps to collect a consistent snapshot, reinstall the same Edge version, and validate the restored environment.
+
+{{< c8y-admon-important >}}
+Always back up `/var/lib/rancher/k3s` and `/datahub` together and restore them to their original paths, ownership, and permissions. Mixing versions or omitting directories can corrupt the cluster.
+{{< /c8y-admon-important >}}
+
+---
+
+### Step 1 - Understand what must be protected
+* `/var/lib/rancher/k3s`: cluster state, objects, certificates, and etcd/sqlite internals
+* `/datahub`: {{< product-c8y-iot >}} DataHub (datalake) content, if provisioned on the node
+
+
+---
+
+### Step 2 - Prepare the Edge node
+Stopping workloads is optional but prevents partial state during the archive.
+
+```shell
+systemctl stop k3s
+```
+
+Ensure only one backup job runs at a time and confirm the filesystem has enough free space to hold the archive.
+
+---
+
+### Step 3 - Create a consolidated backup 
+Run the following command as `root` to collect both directories into a single tarball (or) as per users choice of backup strategy:
+
+```shell
+tar -czvf edge-backup.tar.gz /var/lib/rancher/k3s /datahub
+```
+
+Move `edge-backup.tar.gz` to offline storage or to your backup repository. Retain checksum information if you wish to valdiate prior to recovery.
+
+---
+
+### Step 4 - Prepare the restore target
+1. Install the same operating system (or compatible base image) that originally hosted Edge.
+2. Make sure no prior K3s installation or Edge data exists on the target disk.
+3. Transfer `edge-backup.tar.gz` to the target node.
+
+{{< c8y-admon-caution >}}
+Installing a different Edge or K3s version on top of a restored data set is unsupported and may fail the upgrade guard rails.
+{{< /c8y-admon-caution >}}
+
+---
+
+### Step 5 - Restore the data directories
+Extract the archive to the root of the filesystem so that both directories land in their original locations:
+
+```shell
+tar -xzvf edge-backup.tar.gz -C /
+```
+
+Confirm the directories exist and contain the expected ownership:
+
+```
+/var/lib/rancher/k3s
+/datahub
+```
+
+---
+
+### Step 6 - Reinstall the matching Edge release
+Re-run the installer with the **exact version** captured in the backup:
+
+```shell
+c8yedge install --version <previous_version>
+```
+
+The installer reattaches the restored cluster state and Edge-specific configuration.
+
+---
+
+### Step 7 - Verify health
+After the services start, validate the cluster and workloads:
+
+```shell
+kubectl get nodes
+kubectl get pods -A
+```
+
+Ensure the {{< product-c8y-iot >}} Edge UI loads and that critical applications are functional before handing the system back to operations.
+
+---
+