The ImageMetricsITCase failing with RH Mandrel because postgres is bigger than expected

[jedla97]

Describe the bug

The ImageMetricsITCase failing with RH mandrel (registry.access.redhat.com/quarkus/mandrel-23-rhel8:23.0) as it's bigger then then expected compared to the upstream.

The issue is same as #37882 . Tha new version was released 11 days ago 23.0-19 and it's start failing. It seems that it is just little bigger than before.

Graalvm issue from previous issue is graalvm/mandrel#654 (just for the link the issues)

Expected behavior

No response

Actual behavior

Test is failing with this output:
Expected image_details.total_bytes to be within range [79228920 +- 3%] but was 81620672 ==> expected: <true> but was: <false>

For info in December it was 81149928 now it's 81620672

How to Reproduce?

Failing test

mvn -f integration-tests/pom.xml clean dependency:tree verify -Pnative -Dquarkus.native.container-build=true -Dquarkus.native.native-image-xmx=4g -U -V -B -fae -Denforcer.skip=true -Dtest-containers -Dstart-containers -Dquarkus.native.builder-image=registry.access.redhat.com/quarkus/mandrel-23-rhel8:23.0-19 -pl jpa-postgresql -Dpostgres.image=postgres:14

The -Dquarkus.native.builder-image=registry.access.redhat.com/quarkus/mandrel-23-rhel8:23.0-19 can be substitute as -Dquarkus.native.builder-image=registry.access.redhat.com/quarkus/mandrel-23-rhel8:23.0

Passing test

mvn -f integration-tests/pom.xml clean dependency:tree verify -Pnative -Dquarkus.native.container-build=true -Dquarkus.native.native-image-xmx=4g -U -V -B -fae -Denforcer.skip=true -Dtest-containers -Dstart-containers -Dquarkus.native.builder-image=registry.access.redhat.com/quarkus/mandrel-23-rhel8:23.0-16 -pl jpa-postgresql -Dpostgres.image=postgres:14

Output of uname -a or ver

Having it on RHEL 8.9 and Fedora 39

Output of java -version

No response

Mandrel or GraalVM version (if different from Java)

No response

Quarkus version or git rev

No response

Build tool (ie. output of mvnw --version or gradlew --version)

No response

Additional information

No response

[quarkus-bot]

/cc @Karm (mandrel), @ebullient (metrics), @galderz (mandrel), @jmartisk (metrics), @zakkak (mandrel)

[zakkak]

It is actually failing with both 23.0-26 and 23.0-29 on Quarkus main for me:

23.0-16

Expected image_details.total_bytes to be within range [79228920 +- 3%] but was 81637352 ==> expected: <true> but was: <false>

23.0-19

Expected image_details.total_bytes to be within range [79228920 +- 3%] but was 81653440 ==> expected: <true> but was: <false>

The difference between 23.0-16 and 23.0-19 is ~0.02%

Upstream 23.0.3.0-Final-java17-2024-02-04

Building with the latest upstream 23.0 builder image results in a binary with a total size of 79727808 bytes (which is ~0.6% higher than the baseline).

Upstream 23.0.2.1-Final-java17-2024-01-21

Building with the previous upstream 23.0 builder image results in a binary with a total size of 79712248 bytes, so the difference between the upstream versions is again ~0.02%.

Conclusion

Since the differences are reasonable and there is an increase both in upstream and the Red Hat builds of Mandrel, I suggest we raise the baseline (to 79712248) for 23.0 (I will open a PR).