Closed fredsig closed 1 week ago
Thanks for unblocking the CI @guilload. I can see it failed lint as I didn't bump the version of this chart. If you are ok with this PR and it gets approved, what value you recommend to be changed? Let me know if there's something I can do. Thank you.
LGTM. Thanks for linking the blog post. See also this ALB controller issue. Can you please bump the chart's version (patch) so we can merge this PR?
Yep, this is exactly the original issue. I've bumped the Chart version to 0.7.2. Thank you!
Introduced pod lifecycle. When using the AWS Load Balancer Controller, deployments of pods trigger 5xx gateway errors when doing config changes, since the target group registration/deregistration of pods is out of sync with kubernetes service network orchestration (when an indexer pod gets terminated with a SIGTERM, it's done immediately, not giving enough time to drain connections from the ALB's target group, triggering 5xx errors to the client). Introducing a sleep time on indexers can help avoid this during pod termination (as a pre-stop action).
Related: https://medium.com/@imprintpayments/mastering-the-challenges-of-using-alb-ingress-in-kubernetes-8c28a8f826c5