Is there any configuration snapshot mechanism on kubernetes?
The goal is to take a snapshot of all deployments/services/config-maps etc and apply them to a kubernetes cluster.
The steps that should be taken.
Take a configuration snapshot
Delete the cluster
Create a new cluster
Apply the configuration snapshot to the new cluster
New cluster works like the old one
These are the 3 that spring to mind, with kubed being, at least according to their readme, the closest to your stated goals:
Ark
kubed
kube-backup
I run Ark in my cluster, but (to my discredit) I have not yet attempted to do a D.R. drill using it; I only checked that it is, in fact, making config backups.
State of the kubernetes is stored on etcd, so back up etcd data and restore would be able to restore cluster. But this would not backup any information stored in persistent volumes, that needs to be handled separately.
backup operater provided by coreos is a good option:
https://coreos.com/operators/etcd/docs/latest/user/walkthrough/backup-operator.html
Taking backups with etcdctl :
https://kubernetes.io/docs/tasks/administer-cluster/configure-upgrade-etcd/
https://github.com/coreos/etcd/blob/master/etcdctl/README.md
Heptio ark has capability to backup config and also volumes :
https://github.com/heptio/ark
if you want a UI based option, these would be good :
https://github.com/kaptaind/kaptaind
https://github.com/mhausenblas/reshifter
Related
There is no clear information about how to make a backup and restore from a regular node like node01 for instance, I mean:
Operating etcd clusters for Kubernetes shows information like how to use it and
ETCD - backup and restore management shows some of the necessary steps.
But how about in the cert exam, you are operating most of the time from a regular node01, the config files are not the same? Can some one elaborate?
Thanks
It is impossible to backup cluster from a regular node using etcd. The etcd can only be run on a master node.
But you can backup your Kubernetes cluster by command: etcdctl backup. Here you can find completely guide, how to use etcdctl backup command.
Another way is making a snapshot of your cluster by command: etcdctl snapshot save.
This command will let you create incremental backup.
Incremental backup of etcd, where full snapshot is taken first and then we apply watch and persist the logs accumulated over certain period to snapshot store. Restore process, restores from the full snapshot, start the embedded etcd and apply the logged events one by one.
You can find more about incremental backup function here.
What can be done to backup kubernetes PVC regularly for GCP and AWS?
GCP has VolumeSnapshot but I'm not sure how to schedule it, like every hour or every day.
I also tried Gemini/fairwinds but I get the following error when for GCP. I installed the charts as mentioned in README.MD and I can't find anyone else encountering the same error.
error: unable to recognize "backup-test.yml": no matches for kind "SnapshotGroup" in version "gemini.fairwinds.com/v1beta1"
You can implement Velero, which gives you tools to back up and restore your Kubernetes cluster resources and persistent volumes.
Unfortunately, Velero only allows you to backup & restore PV, not PVCs.
Velero’s restic integration backs up data from volumes by accessing the node’s filesystem, on which the pod is running. For this reason, restic integration can only backup volumes that are mounted by a pod and not directly from the PVC.
Might wanna look into stash.run
Agree with #hdhruna - Velero is really the most popular tool for doing that task.
However, you can also try miracle2k/k8s-snapshots
Automatic Volume Snapshots on Kubernetes
How is it useful? Simply add
an annotation to your PersistentVolume or PersistentVolumeClaim
resources, and let this tool create and expire snapshots according to
your specifications.
Supported Environments:
Google Compute Engine disks,
AWS EBS disks.
I evaluated multiple solutions including k8s CSI VolumeSnapshots, https://stash.run/, https://github.com/miracle2k/k8s-snapshots and CGP disks snapshots.
The best one in my opinion, is using k8s native implementation of snapshots via CSI driver, that is if you have a cluster version > = 1.17. This allows snapshoting volumes while in use, doesn't require having a read many or write many volume like stash.
I chose gemini by fairwinds also to automate backup creation and deletion and restoration and it works like a charm.
I believe your problem is caused by that missing CRD from gemini in your cluster. Verify that the CRD is installed correctly and also that the version installed is indeed the version you are trying to use.
My installation went flawlessly using their install guide with Helm.
I want to deploy a etcd cluster using a Helm Chart with the possibility to restore in case of a cluster-level failure.
https://github.com/helm/charts/tree/master/incubator/etcd
Deployment using the following Helm Chart works fine - I can use it with success as a storage backend for my application, the problem is - I can't restore it from a snapshot. Tried to restore it, but every time a spin up a new cluster, exec into it and try to restore the snapshot (copied into the pod), the data is not restored. Or in other words - I see that the data folder is re-created inside the pod, but still, no data is inserted, looks like etcd is still using the "default.etcd" data folder, which is mounted as the default PVC, instead of a "restored.etcd".
Any one had similar issues with restoring an etcd helm chart data on a k8s cluster? Any tips?
Is it possible to undo "gcloud container clusters delete" command?
Unfortunately not: Deleting a Cluster
All the source volumes and data (that are not persistent) are removed, and unless you made a conscious choice to take a backup of the cluster, it would be a permanent operation.
If a backup does exist, it would be a restore from backup rather than a revert on the delete command.
I suggest reading a bit more into the Administration of a cluster on Gcloud for more info: Administration of Clusters Overview
Unfortunately if you will delete cluster it is impossible to undo this.
In the GCP documentation you can check what will be deleted after gcloud container clusters delete and what will remain after this command.
One of the things which will remain is Persistent disk volumes. It means that if your ClaimPolicy was set to Retain and your PV status is Released you will be able to get data from PersistentVolume. To do that you will have to create PersistentVolumeClain. More info about ReclaimPolicyhere.
Run $ kubectl get pv to check if it is still bound and check ReclaimPolicy. Similar case can be found in this github thread.
In this documentation you can find step by stop how to connect pod to specific PV.
In addition, please note that you can backup your cluster. To do this you can use for example Ark.
I am working with Kube-Aws by coreos to generate a cloud formation script and deploy it as part of my stack,
I would like to upgrade my kubernetes cluster to a newer version.
I don't mind creating a new cluster, but what I do mind is recreating all the deployments/services etc...
Is there any way to take the configuration and replace/transfer them to the new cluster? maybe copy the entire etcd data? will that help?
Use kubectl get --export=true on all the resources that you want to move into a new cluster and then restore them that way.
kubectl get <pods,services,deployments,whatever> --export=true --all-namespaces=true