Routing with NLB Elastic IP problem, listener malfunctioning

Our environment is set with ALB and NLB, which have 2 Elastic IPs set. On last Thursday some of our clients started to experience inability to reach our web application or extreme slowness in its work. We didn't find any errors in logs or failure in health checks. At some point we did a manual health check for each of the Elastic IP DNS addresses and we found that one of them didn't respond. That explained why the problem happened only for some of our customers - the address remained cashed in their system. What seemed to resolve the problem was to delete and create listeners. What we still don't understand is what has created the problem, there were no changes in any NLB, ALB, Target group or Elastic IP settings for the last 3 month. And why there wasn't failure in the health checks.