[GH] (shardingsphere/fix-issue-39765-otel-span-scope): Workflow run "CI - GraalVM" failed!

[GitBox <gi...@apache.org>]

The GitHub Actions job "CI - GraalVM" on shardingsphere.git/fix-issue-39765-otel-span-scope has failed.
Run started by GitHub user HhH666-ux (triggered by HhH666-ux).

Head commit for run:
df6d6aebd51a6af3c1751352f38889e8691ad237 / HhH666-ux <32...@qq.com>
Reclaim abandoned JDBC spans on a deadline instead of on map access

The static registry used a WeakHashMap, so an entry was only expunged by a
later call on the map, and the value kept the key alive anyway. Between
beforeMethod and afterMethod the plugin can be disabled, and then neither
onThrowing nor afterMethod runs, so the span and its SQL attributes stayed
retained until unrelated tracing traffic touched the map again.

Register each span in a ConcurrentHashMap together with its creation time and
let a daemon scheduler end and release entries that are older than a generous
retention bound. Reclamation no longer depends on garbage collection policy,
on later SQL execution or on map access, and the advice keeps reporting the
span instead of dropping it.

The test now drives the reclaimer through its deadline, so it no longer
depends on System.gc(), and it reads the field with Plugins.getMemberAccessor()
instead of setAccessible. endSpan and findSpanByStatement follow their first
callers, and the explanatory line comments moved into JAVADOC.

Report URL: https://github.com/apache/shardingsphere/actions/runs/40516234651

With regards,
GitHub Actions via GitBox