Auto Suspension

This page describes about the auto suspension feature in the cluster.

Auto suspension suspends/turns off a cluster when kept unoccupied/free for a specified period in minutes. Depending on the nature of your workloads, your instance may be in a state of rest during off-peak times, increasing the cluster compute cost. Auto-suspension allows you to reduce the cost of compute during these times.

In some cases, not all executors will be available immediately when a cluster is created due to infrastructure availability limitations. The remaining executors will become active as soon as resources are available.

Therefore, the auto-suspension time is calculated starting from the time all the executors in a cluster come online.

Auto Suspension Time Adjustment

When a user edits the auto-suspension time, the new value will be calculated based on the current active session's start time, and the action will be taken accordingly.

Examples:

1. Reducing Auto-Suspension Time:

   • If the current auto-suspension time is 30 minutes and the user updates it to 5 minutes, and the cluster has already been active for 5 minutes, the cluster will be suspended immediately.

2. Increasing Auto-Suspension Time:

   • Suppose the current auto-suspension time is 30 minutes and the user updates it to 50 minutes, and the cluster has already been active for 20 minutes. In that case, the cluster will be suspended after an additional 30 minutes (i.e., 50 minutes minus the active session time of 20 minutes).