AWS Load balancer keeps draining sakai instance

[adnan ashraf]

Hi,

I am running sakai in AWS ECS Fargate cluster. Everything works fine, its just that the load balancer keeps draining sakai instance because it thinks sakai is unhealthy target in LB target group.

initially I had health check at root "/", then changed to "/library/skin/tool_base.css"(which always returns 200 ok in browser). I looked through all the conversation here in this forum, chatGPT and googled it as well but not finding a simple way to make this work.

I will appreciate if someone can share how they are adding health check to keep target group healthy, thanks

Adnan

CTO at DevTek

[Adam Marshall]

I’m about as far from an expert on this matter as it is possible to be, but is happening because of the way Java manages its memory usage – from outside, it always looks like all available memory is being used? This isn’t actually a problem but the load balancer thinks it is?

 

Maybe!

 

adam

--
You received this message because you are subscribed to the Google Groups "Sakai Development" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sakai-dev+...@apereo.org.
To view this discussion visit https://groups.google.com/a/apereo.org/d/msgid/sakai-dev/d03d1f32-779e-4405-9565-342264bddf3dn%40apereo.org.

[Sam Ottenhoff]

 it thinks sakai is unhealthy target in LB target group.

Is the target Sakai unhealthy? What happens when you visit it manually after an alert? 

I usually recommend failing a target node after three successive failures spaced 20 seconds apart.

Optimizing Sakai is largely about JVM memory management and GC performance. There are good threads from Earle in the archives about best settings for GC.