cfml

image.png

I'm deploying Lucee on docker and running on EKS. I have the JVM memory args set to `-Xms2048m -Xmx2048m` (2GB). But it appears that when the container/Lucee starts up it uses around 4GB and then drops. Shouldn't the JVM args limit the memory available to Lucee/Java? Is that normal for high memory allocation on startup?

Your operating system will sometimes report more memory used than your heap is

hmm...in EKS I'm setting the memor limit to 2GB
```          resources:
            limits:
              cpu: 1000m
              memory: 2160Mi
            requests:
              cpu: 500m
              memory: 2160Mi```
but I think that's for the container (OS+Lucee) if I want to give JVM 2GB then I should probably increase that limit to 4GB (because that's apparantly what's being used on startup)?

Smaller Lucee instances on 5.4 official images run swimmingly at 1GB.  Most of my customer's apps run at 2GB.

But I expect what Brad is saying is correct -- the non-heap memory is probably being misreported.

We’re using `lucee/lucee:5.4.5.15-SNAPSHOT-light-nginx-tomcat9.0-jre11-temurin-jammy` image.

you could probably save some memory by not having nginx in that image and using an http proxy like traefik instead (or using the AWS load balancer).  I'm personally not a fan of having the http server in the same image since it breaks the "one service. one function" rule.