Before reporting an issue

• I have searched existing issues
• I have reproduced the issue with the latest release

Area

core

Describe the bug

CI job running with Keycloak Nightly images discovered a 12-14% increase in startup time, augmentation time, and memory footprint after start between 2023-04-28 and 2023-04-29 (both at 2:25 AM UTC). The regression has been present in all subsequent runs since then.

Version

nightly

Expected behavior

Server startup time and augmentation time (as reported in the server log) and memory footprint after start at the level of pre-Quarkus 3 merge baseline.

Actual behavior

+12 .. 14% regression compared to baseline.

How to Reproduce?

Measure startup time, augmentation time as reported in Keycloak server's pod log; memory usage as reported in /sys/fs/cgroup/memory/memory.usage_in_bytes or /proc/1/smaps inside the Keycloak pod. Utility scripts for this will be added to the keycloak-benchmark project shortly.

Anything else?

No response