-
Notifications
You must be signed in to change notification settings - Fork 273
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Can't start kine backend database as static pod to replace etcd #7332
Comments
Don't do that. As I said when you asked on Slack:
|
SUSE and Synadia are working on a two Master HA Kubernetes Setup for the Edge. They are doing that on two hosts in edge environments, where you - as possible Use Case - only have two fire zones. If one Fire Zone is down, the Kubernetes Cluster will be down without such solutions as described here. I“m sorry that you closed this issue. In my opinion, the behaviour of Kine backed clusters is different as for ETCD backed clusters and this shouldn‘t be the case. |
We are not planning on supporting external datastore running IN the cluster as static pods, regardless of the datastore type. We are able to do that with etcd because we manage its configuration and are able to ensure that it comes up in the correct sequence - but this is not easy to accomplish with arbitrary 3rd party databases. I will also note that kine on NATS does not currently pass kubernetes conformance tests. Ref: k3s-io/kine#321 |
Hi,
I try to start a mariadb static pod as a kine backend to replace etcd.
If I start mariadb with systemd and create a etcd database. Everything works as expected.
If I start a static pod by creating a mariadb pod yaml in
/var/lib/rancher/rke2/agent/pod-manifests
and start the rke2-server systemd service, I get an immediate error in the journal logs, that nothing is listening on port 3306.
I would expect, that RKE2 starts all static pods (also MariaDB) and AFTERWARDS tries to connect to ETCD -> kine (and therefore also to MariaDB) plus: ETCD on its own is also started as a static pod.
Is that currently not possible?
Thanks and greetings,
Josef
The text was updated successfully, but these errors were encountered: