The balancer emit a log message every time it fails to find a destination shard to move out chunks from a draining shard.

The problem is that the balancer emits this log warning for every chunk to be drained. This means that potentially in one balnacer round (few secs) we could emit millions of logs leading the machine out of disk space.

Another issue is that we should emit a warning only if we actually have some shard still available in this round (not already used in this balancing round) to receive chunks.
If all the shards available are already in draining mode, it is expected that we can't find a recipient.

I believe the same happens also in case no recipient can be found while enforcing zones.