K8Syncer is a tool to backup k8s resources into some kind of storage. While it was designed to be easily extensible with further storage options, the primary goal was to be able to backup resources from a k8s cluster into a git repository, and the tool does not support much more than this use-case at the moment.
See here for a documentation of the supported storage types and here for how to configure the tool.
At the moment, K8Syncer reacts only to changes on the metadata fields generation, labels, and ownerReferences of k8s resources. This means that resources kinds for which the k8s apiserver does not increase the generation are not synced on a spec update. This applies to namespaces and secrets, for example. A workaround would be to trigger a sync by changing a label on the resource.
Although it is possible to run K8Syncer locally - one simply has to provide its configuration via --config and a kubeconfig for the target cluster, either via KUBECONFIG env var or --kubeconfig - it was designed to run as a controller inside a kubernetes cluster. The easiest way to install it is by using the provided helm chart.
image: # can usually be left out
repository: eu.gcr.io/gardener-project/k8syncer
tag: "0.1.0"
config:
# kubeconfig has to be provided if k8syncer should watch another cluster than the one it is running in
# kubeconfig: |
# apiVersion: v1
# clusters:
# - cluster: ...
syncConfigs: # goes directly into the k8syncer configuration, see docs/usage/configuration.md
- id: dummyWatcher
resource:
group: k8syncer.gardener.cloud
version: v1
kind: Dummy
storageRefs:
- name: mockStorage
state:
type: annotation
verbosity: detail
storageDefinitions: # goes directly into the k8syncer configuration, see docs/usage/configuration.md
- name: mockStorage
type: mock
# resources:
# requests:
# cpu: 100m
# memory: 256Mi
# limits:
# cpu: 500m
# memory: 2Gi
logging:
verbosity: info # error, info, or debug (defaults to 'info' if omitted)Have a look at the documentation: