Retries in cluster requests

When cluster request to object storage is executed, some node can be failed (swarm downscale, some kubernetes actions, etc.)

Now executed request is also failed.

Need to reuse other alive nodes to continue request execution.