Describe the feature:

A rules to monitor segment counts at shard/index/node levels and alert when the number of segments becomes excessive for each context.

Describe a specific use case for the feature:

A rule to alert on excessive segment counts that might indicate problems with successful merging affecting different cluster contexts.

Problems this might help avoid or alert on may include:

• impact from bugs, eg Use local segment fieldInfos to lookup tsdb merge stats elasticsearch#132597
• impact from resource over utilisation (eg too many indices/segments on a node)
• best practice not being followed or edge cases that may lead to performance or stability impacts
  • lack of force merge for a use case where it's generally required
  • excessive scrolls holding segments up from merging out deleted lucene docs, leading to excessive segment counts
  • extreme update workloads where merge is throttling or just isn't keeping up for particular reasons leading to unhealth segment counts