By using AWS re:Post, you agree to the AWS re:Post Terms of Use
AWS re:Postre:Post

Classic Load Balancer terminates large downloads mid request

0

Over the past week, clients have reported that any downloads approximately 600+ MB will be terminated mid request within a few seconds of the original request starting. I've checked multiple places to make sure its not my app that's dropping the connection or the EC2 instance and have verified that instances outside the Classic ELB can download the same file just fine.

My production workload is in Opsworks Stacks and I bring up and take down the webservers according to load, and time of day. Opsworks very cleanly adds the instances to the load balancer after they come up and removes the instance from the load balancer before shutting down. I have looked into migrating to NLB or ALB, however since AWS still does not support those load balancers in Opsworks Stacks I am stuck using the Classic Load Balancer.

After pounding my head against the wall I opened a support ticket to report a possible problem. AWS Support confirms the bug is known and "Classic Load Balancer experiences high resource utilization, when large files are being uploaded and downloaded to or from the target EC2 instances". Also I was informed there is no ETA of a fix. This was April 11, I have escalated to my account manager as well, but apparently he has not been able to get any traction either.

I am posting so there is visibility.

Topics
Networking & Content Delivery
Tags
Elastic Load BalancingClassic Load Balancer
Language
English
jwestbrook
asked 2 years ago1278 views
1 Answer
0

Hello, and thank you for reaching out!

I have reviewed the case you opened with AWS Support and the resulting investigation from the internal ELB team regarding this known issue with Classic Load Balancers.

This does appear to be a unique situation wherein the recommended workaround of migrating to ALB / NLB is explicitly unsupported by the AWS OpsWorks Stacks service utilizing ELB. I do see that there some environments wherein an Application Load Balancer can be used with some OpsWorks Stacks, but I do understand if this workaround does not fit your environment.

I wanted to let you know that I have engaged internal teams further to determine if there is more information or a workaround that can be found for this issue. I encourage you to follow up on this via the original support case or via your account manager.

AWS
SUPPORT ENGINEER
Justin_B
answered 2 years ago
  • HenryChen
    12 days ago

    Hi Justin, I met this issue recently. Our application is basing on EKS, the service is exposed by LoadBalancer with k8s service. Any suggestion for the ELB configuration for that ?

You are not logged in. Log in to post an answer.

A good answer clearly answers the question and provides constructive feedback and encourages professional growth in the question asker.

Guidelines for Answering Questions

Relevant content