You no longer need to stop the virtual machine just to take a snapshot. Kubernetes clusters are able now to automatically handle master node failure in the high availability mode.
Online snapshotting is now available
You can now create snapshots at any moment, regardless of whether the disk is attached or detached, and the virtual machine is stopped or running. The only limitation is that you can take no more than 16 snapshots at a time for all projects and users that you have created as a customer.
A point-in-time snapshot is taken immediately and does not affect read/write operations – the disk can be attached, detached, or even deleted without waiting for the snapshot to be transferred to the object storage. You can create snapshots as often as needed, but the next snapshot of the same disk can only be taken after the upload is complete.
In addition, in one of the upcoming CROC Cloud releases it will also be possible to create templates for virtual machines without stopping them.
In the fault-tolerant cluster mode the failed master node is now restarted automatically. The entire recovery process takes a few minutes, yet the cluster remains fully functional.
When you create a Kubernetes cluster a required security group is added. It contains two rules to enable traffic from and within the cluster to ensure its correct operation.
If you add or delete worker nodes, and there are not enough resources, a required template cannot be found, or a worker node cannot be deleted, the cluster is reconfigured to the extent possible. In this case, the cloud will generate an error message, it can be found in the "Has warnings" column in the cluster table and on the "Warnings" tab on the cluster page.
Other new features
If you want to know how many objects are in a bucket, what is the bucket size, and who is a bucket owner, you can find these data on the Information tab on the bucket page.