Preemptible node groups
In a preemptible node group, each node runs for a maximum of 24 hours. During this time, nodes can be stopped by Selectel at any time — for example, if the virtual host does not have enough resources for other clusters.
The recovery will start immediately after the node is stopped. If free resources are sufficient for recovery, the duration of node recovery will depend on the type of boot disk:
- a network boot disk node is recovered in less than one minute;
- node with local boot disk is restored up to 10 minutes.
Data recovery on the node also depends on the type of boot disk. If a network boot disk is used, all data is saved, if a local boot disk is used, all data is deleted.
After each recovery, the 24-hour countdown starts over.
Preemptible node groups support all features that are available for regular node groups, while the cost of resources in such groups is 70% lower on average.
You can make the node group preemptible in the control panel when the cluster creation or adding a node group. Any node group can be preemptible configurations.
Limitations
Preemptible node groups are only available in clusters on cloud servers in the pool segments ru-7a and ru-7b.
We do not guarantee the level of availability as with normal node groups — preemptible node groups are not affected Kubernetes cluster SLA.
Cost
The resource cost of an interruptible node group is on average 70% lower than the resource cost of a normal node group with the same configuration.
During operation, resources in the interrupted node group are charged at the Cloud platform resource payment models.
After the interruption:
- vCPU, RAM, GPU, local disks are no longer charged starting from the next hour after the stop;
- network disks continue to be charged.