Using init-container, the AMQ pod has Xmx value equals to 1/2 resources.requests.memory, as defined in the Activemqartemis CR at deploy time. There is no way to change this value by updating the CR after initial deploy (see ENTMQBR-5748).

In this example, I'm initially setting requests==limits with 1 CPU and 1GiB of memory. After the initial deployment, I directly edit the broker STS in order to raise resources to 3 CPU and 3GiB of memory (workaround).

Non-terminated Pods:                      (17 in total)
  Namespace                               Name                               CPU Requests  CPU Limits  Memory Requests  Memory Limits  Age
  ---------                               ----                               ------------  ----------  ---------------  -------------  ---
  broker                                  my-broker-ss-0                     3 (19%)       3 (19%)     3Gi (4%)         3Gi (4%)       9m37s
  broker                                  my-broker-ss-1                     3 (19%)       3 (19%)     3Gi (4%)         3Gi (4%)       10m

As you can see, the workaround works, but when I check broker logs I still have the global-max-size value computed from the deployment value (256 MiB, where Xmx == 1/2 deployment mem == 512 MiB):

2021-11-25 12:26:01,561 INFO  [org.apache.activemq.artemis.core.server] AMQ221057: Global Max Size is being adjusted to 1/2 of the JVM max size (-Xmx). being defined as 268,435,456

This makes impossible to recover from a situation where you have a huge backlog of messages and the broker needs to load the whole journal into memory at startup.

A workaround here would be to use init-container to overwrite the global-max-size value, but this is not something OLM/GUI users are used to do.