[Release-1.26] - --etcd-disable-snapshots flag usage leads to unexpected behavior

[vitorsavian]

Backport fix for --etcd-disable-snapshots flag usage leads to unexpected behavior

• --etcd-disable-snapshots flag usage leads to unexpected behavior #8031

[ShylajaDevadiga]

Validated on release-1.26 branch using commit id

Environment Details

Infrastructure
Cloud EC2 instance

Node(s) CPU architecture, OS, and Version:

$ cat /etc/os-release 
NAME="Ubuntu"
VERSION="20.04 LTS (Focal Fossa)"

Cluster Configuration:
Single node

Config.yaml:

$ cat /etc/rancher/k3s/config.yaml 
cluster-init: true
etcd-disable-snapshots: true
etcd-s3: true
etcd-s3-access-key: 'KEY'
etcd-s3-secret-key: 'SECRET'
etcd-s3-region: 'REGION'
etcd-s3-bucket: 'BUCKET'

Steps to reproduce

Copy config.yaml
Install k3s
Take a snapshot passing s3 params
Stop k3s service
Restore the snapshot passing s3 params
Start k3s service
Verify the cluster is up and running

Replication results:
On k3s version v1.26.6+k3s1

Restore fails with path does not exist error

FATA[0001] starting kubernetes: preparing server: start managed database: etcd: snapshot path does not exist: on-demand-ip-172-31-15-87-1692378660 

Validation results:

INFO[0014] Managed etcd cluster membership has been reset, restart without --cluster-reset flag now. Backup and delete ${datadir}/server/db on each peer etcd server and rejoin the nodes 

$ k3s -v
k3s version v1.26.7+k3s-fdc7485c (fdc7485c)
go version go1.20.6

$ kubectl get nodes
NAME              STATUS   ROLES                       AGE   VERSION
ip-172-31-6-153   Ready    control-plane,etcd,master   55m   v1.26.7+k3s-fdc7485c