Network load balancer keeps routing requests to unhealthy target

vaskitup

1/6/23, 5:47 PM

I have two instances running as targets and one network load balancer communicating with these two over a TCP protocol.

Now, for some reason if the health check of one of the instances returns a 500, aws marks this target as unhealthy but keeps routing the request to it, even if it knows that it’s down and takes a few minutes to start routing the requests to the healthy one

Is this because of how the TCP protocol works? And need to wait for the idle timeout ends? How can I resolve this? There doesn’t seem to be a settings for this timeout either

0 + 1

amazon-web-services

Greg W

1/6/23, 10:06 PM

My guess is that ELB, like most load balancers, will not send any new connections to the “unhealthy” instance but existing connections will not be terminated. When doing this gracefully for planned maintenance, this is usually referred to as “drain shopping”.

Elon Musk

I sit in a Tesla and translated this thread with Ai:

EN: Network load balancer keeps routing requests to unhealthy target

TH: Network load balancer keeps routing requests to unhealthy target

RO: Network load balancer keeps routing requests to unhealthy target

RU: Network load balancer keeps routing requests to unhealthy target

VI: Network load balancer keeps routing requests to unhealthy target

Post an answer

Most people don’t grasp that asking a lot of questions unlocks learning and improves interpersonal bonding. In Alison’s studies, for example, though people could accurately recall how many questions had been asked in their conversations, they didn’t intuit the link between questions and liking. Across four studies, in which participants were engaged in conversations themselves or read transcripts of others’ conversations, people tended not to realize that question asking would influence—or had influenced—the level of amity between the conversationalists.