I am a bit confused about Rancher’s definition of HA.
First, let me explain my setup for Rancher HA:
I followed the Rancher HA Install for Air Gap Environment using the RKE Add-on instructions
I am using rancher:v2.0.8 and kubernetes v1.11.2.
I am using 3 ec2-instances for my infrastructure and use an ELB to front the 3 instances. The ELB is setup following the NLB configuraiton document
I terminate TLS on the ingress-controller using certs from file and I correctly configure my truststore for the CA
Ultimately, I was able to get the cluster booted up properly ONLY when there was one rancher container deployed to the local k8s cluster. How is this Highly Available? I need rancher highly-available because I call its REST API for cluster-autoscaling (using custom node cluster built on ec2 infrastructure) as well as use the REST API for helm chart deployments.
Is there something I am missing from the install documents? I assumed that the replica count for rancher should be updated to 3 for the HA installation but could never get passed the logging error on the cattle-agents which said:
"Failed to connect to proxy" error="websocket: bad handshake"
If an AZ goes down, I want to have another rancher container available in another AZ to be able to continue operations.
There is a banner present on both referenced pages:
Important: RKE add-on install is only supported up to Rancher v2.0.8
Please use the Rancher helm chart to install HA Rancher. For details, see the HA Install - Installation Outline.
If you are currently using the RKE add-on install method, see Migrating from an HA RKE Add-on Install for details on how to move to using the helm chart.
Doesn’t that banner mean that I can used the RKE add-on install since I use Rancher v2.0.8?
I tried both install instructions and found it much easier to test Rancher Cluster bootup with the RKE add-on install instructions.
I also received those same errors when using a helm chart deployment… What in particular is different about the deployment via helm chart vs the deployment via rke add-on?
Also, keep in mind that I need to use the Air Gap install so I need to download the helm chart via github first, and bring it into my environment
Support for running multiple pods was added in Rancher v2.1.0 and is only compatible when installed using the Helm chart:
Rancher HA improvements - Administrators can now configure an HA setup with more than one Rancher server. This provides better redundancy and horizontal scalability.
Getting “the same error” is a bit too generic, if you have reproducible steps, please file an issue on GitHub, using the issue template, describing the minimal steps to reproduce and the outcome as described on https://rancher.com/docs/rancher/v2.x/en/contributing/#bugs-issues-or-questions. It also helps to provide full logging as it helps determining what happens before and after a certain event.
Are you saying that it is not possible to deploy multiple rancher pods in HA configuration when using rancher v2.0?
Is it only possible to use MULTIPLE rancher containers managing the SAME k8s clusters when using Rancher v2.1?
If this is the case, then my github issue can be closed with the resolution that I must upgrade to v2.1 in order to deploy multiple replicas of rancher in HA configuraiton
I have an issue opened on github but I think attaching the source is automatically flagging this post…
Are you saying that it is not possible to deploy multiple rancher pods in HA configuration when using rancher v2.0?
Is it only possible to use MULTIPLE rancher containers managing the SAME k8s clusters when using Rancher v2.1?
If this is the case, then my github issue can be closed with the resolution that I must upgrade to v2.1 in order to deploy multiple replicas of rancher in HA configuraiton