Better handling of hdd/hdd controller failures

Detail name	Value
Changelog Number	8976
Type	Improvement
Status	Resolved
Fix Versions	Exasol 8.0.1
Resolution Date	2022-05-06

Problem

In some circumstances when a hard drive or raid controller fails on a node, the node can become unresponsive without being able to report an error because essential services are blocked writing to the affected disk, while at the same time remaining online in the cluster because it is still theoretically reachable for the other nodes.

Solution

We intend to add a watchdog that will detect this failure case and remove the unresponsive node, triggering the normal fail-safety mechanics and allowing the rest of the cluster to run unobstructed.