Resize linode with rancher
I just got responsible by an ongoing project that is on linode and uses rancher as container facilitator. I'm learning everything about this right now and I have a few doubts regarding resizing of the host machine.
The architecture consists of one master host that probably contains rancher and 3 clusters. All hosts are full, they have 30GB, linode 2GB. I'm experiencing some issues because of that and I saw that the only option I have is to resize my linodes, for a 4GB or 6GB, because the maxlimit of the linode is already being used.
My question is: will there be any problem if I resize one of the clusters? I mean, will I have to configure rancher again? My containers will stop working after the resizing is finished? I'm afraid that if I resize one of the clusters the live application stops working.
I wouldn't expect you to have any Linode specific issues resizing one of the clusters. Resizing the Linode won't impact your Rancher configuration inherently but, as it involves downtime, rebooting will clear out any configurations that aren't set to persist through a reboot.
My recommendation is that you backup your Linode prior to resizing your cluster. This allows you to resize the cluster and test that live applications will continue running. If something goes wrong, you can immediately rollback to before the resize to mitigate downtime.
Here are some Rancher specific resources I found to assist you:
Once you know you can resize the cluster, go ahead and decrease the size of your plan.
First verify that your disks are using less space than the new plan provides. You would want to make sure you had room to resize by running the following command:
Once you've confirmed that your internal usage is below the new plan's disk size, you are ready to resize. However, if you are using more space then your new plan you will have to delete any unnecessary files.
You can resize a Linode's disks by powering it down and clicking on the 'Edit disk' link, enter the size in the 'New Size' field. The speed at which the disk resizes is largely dependent on how much data you have on disk, and how fragmented that data is on the disk.
Once the disks are the correct size, you are ready to migrate to a smaller plan in the "Resize" tab of the Linode Manager.
I hope this helps!
Linode Support Team
You can then define Node Templates for each of your different Linode Instance Types (and cluster roles, control-plane or worker). Rancher makes it easy to scale up or down the number of nodes in each Node Template.
In a new cluster deployed with this approach, if you want to resize your nodes (Linodes), add large nodes into the cluster, cordone and drain the smaller nodes, and remove the smaller nodes once all of your deployments have been rescheduled.
When dealing with stateful applications, take advantage of the CSI so that your data can be rescheduled to replacement nodes. If you take advantage of the CSI in this way, High Memory instance types may offer the best value.
More details are in the How to Deploy Kubernetes on Linode with Rancher 2.2 guide, including:
Configuring the Linode CCM (Cloud Controller Manager)
- provisions NodeBalancer for LoadBalancer type Ingresses
- sets Node status and annotations based on the Linode details
Configuring the Linode CSI (Container Storage Interface)
- dynamic volume provisioning with Block Storage
- allows Service and Ingress annotations to trigger Linode DNS Manager record updates