We have a 10 members(16CPU,64GB RAM) search head cluster in the same data center. 3 members are preferred captain and F5 will not forward traffic to these 3 members , and parameter captain_is_adhoc_searchhead is configured on 10 members.
Sometimes , one of the search head's load average exceeds 1 because of CPU or memory overuse, then this search head will be not able to response captain's call in time. This member will launch a captain election, and this member will become the new captain even if it not a preferred captain. The captain election process is not over yet until a member with preferred captain parameter become the captain .
The search head cluster is unstabitily during the captain election, profuse schedule search and alert will be skipped, some critical alert will miss . How to solve this problem, how to prevent a non-preferred captain to be elected as captain ?
