Anyone have a hands on guide about how to implement px-dr in gke ?
Jaimon,
There is not a single guide directly for GKE, however our docs do cover setup for GKE. Below is a list of resources that should explain what you need for GKE. Feel free to post any specific questions you may need clarification on as well.
General DR Topic
Install PX on GKE (Operator or Daemonset)
Async DR (See Google Cloud Sections)
Async DR (See google cloud sections)
Blog with video that uses GKE
Thanks!
- How can i configure Metro DR when the cluster is using internal KVDB?
- is that config is supported ?
@Jaimon Metro DR requires an external KVDB (ETCD). It is not supported with internal KVDB.
I followed the cluster pairing doc and it doesnt seems to work.
Few things may be missing there in the doc
- the need of migrationnamespace - not pre created
my cluster pair output looks
storkctl get clusterpair
No resources found.
Any idea ?
Thanks Alex
I tried with external ETCD
@Jaimon the “migrationnamespace” is a generic namespace where you want to create the clusterpair and run your migration. It can be any namespace you want to migrate resources.
“storkctl” command is then used similar to “kubectl” and you need to specify the namespace:
storkctl get clusterpair -n migrationnamespace
Thanks Alex!
Now i can see the status, and storage in Error state
Events:
Type Reason Age From Message
Warning Error 3m17s (x662 over 113m) stork Error from remote cluster: Cannot create cluster pair with self
You have to generate the clusterpair yaml in the destination cluster so it can use the Kubernetes context from the destination.
Then you create the clusterpair resource in the source cluster.
The error you are seeing means you are generating and creating the clusterpair in the same kubernetes cluster.
Thanks for your swift feedback…
I switched the context to source cluster after generating the clusterpair yaml
Anything else that i can verify to fix the issue ?
You can also see the Cannot create cluster pair with self error if you have specified the IP/Token field in the storage options section of the cluster pair file by accident in a Metro DR setup. Try removing the IP and Token field and give it another try