Configure Cluster/Node

Why this matters: Mavvrik shows a default cost the moment a cluster, node, or volume is onboarded, but that default won't match your actual negotiated rates unless you configure it. This is where you correct that, so the Kubernetes cost dashboard reflects what you're actually paying, not a generic estimate.

How Kubernetes rates work

Rates are hourly. The CPU rate is per core, the GPU rate is per GPU, and the Memory and Storage rates are per GB.

Cluster level (Admin > Accounts > Kubernetes > K8s > Edit cluster)

Status: Enabled or Disabled. Setting a cluster to Disabled immediately stops all future data processing for that cluster. When a cluster is disabled, the panel shows only the Status field.
Show in Main Cost Dashboard: when this toggle is on, the cluster is displayed on the main Cost page. When it is off, the cluster is not displayed there.
Cluster Name, Prometheus URL (read-only), Location, Latitude and Longitude. The location and coordinates place the cluster on the Geo Chart.
Default CPU Rate, Default Memory Rate and Default Storage Rate: applied to the nodes in the cluster.
Default Volume Rate: one rate for each storage class in the cluster (for example standard or standard-rwo).
Reset all nodes and volumes rates to default: select this before saving to apply the cluster rates to every node and volume.

Node level (Edit node): CPU Rate, Memory Rate and Storage Rate for that node. For a node that has GPUs, the panel also shows a GPU Rate for the GPU model on the node (for example Tesla-T4). The GPU rate is the cost of one GPU for one hour. Nodes without GPUs show no GPU values.

Volume level (Edit volume): a rate for that volume.

How node cost is calculated

CPU cost = number of cores × CPU rate
GPU cost = number of GPUs × GPU rate
Memory cost = memory in GB × Memory rate
Storage cost = storage in GB × Storage rate

Example. A node with 2 cores, 3.82 GB of memory and 47.39 GB of storage, with rates of $0.05 per core-hour, $0.005 per GB-hour and $0.0001 per GB-hour, costs about $0.10 + $0.02 + $0.005 = $0.12 per hour. These rates are example values, not defaults.

GPU example. A node has 2 Tesla-T4 GPUs and a GPU rate of $2 per GPU-hour. Its GPU cost is 2 × $2 = $4.00 per hour. The node table shows this in the GPU Cost (Hourly) column, next to the GPU count and GPU model.

Update the rates

Below are the steps to update the cluster

Click on "Edit Cluster" for the cluster you wish to update.
Adjust the default rates as needed.
Select the checkbox for "Reset all nodes and volumes."
Click on "Configure."
After updating the cluster, the cost for all nodes and volumes will be updated simultaneously.
Wait for the next ingestion cycle, scheduled every 6 hours, to see the updated cost on the Kubernetes cost dashboard.

Below are the steps to update the Node

Click on "Edit Node" for the node you wish to update.
Adjust the default rates as needed. On a GPU node, this includes the GPU Rate.
Click "Configure."
Wait for the next ingestion cycle, scheduled every 6 hours, to see the updated cost on the Kubernetes cost dashboard.

Below are the steps to update the Volume

Click on "Edit Volume" for the volume you wish to update.
Adjust the default rates as needed.
Click "Configure."
Wait for the next ingestion cycle, scheduled every 6 hours, to see the updated cost on the Kubernetes cost dashboard.

Update many nodes at once

To change the rates for several nodes in one step, use a rate file:

Click Update Rates above the Nodes table.
Click Rate Template to download the template.
Edit the rates in the file and save it as a .csv file.
Click Rate File, choose your file, and click Update.

The Update Node Rate dialog previews these columns: NODE, CPUS, RAM SIZE, STORAGE SIZE, CPU RATE, RAM RATE and STORAGE RATE. For a cluster with GPUs, it also shows GPUS and GPU RATE.