Today's nightly Docker build failed again:
  https://github.com/apache/solr/actions/runs/35678697209

The failing leg is specifically `branch_9x, 17, full,
eclipse-temurin:21-jre-jammy, -java21` (all other matrix legs -- main,
branch_10x x2, branch_9x's default JDK17 image -- succeeded in the same
run). The actual error is inside the image build, during apt-get install:

#17 43.24 dpkg: error processing package libc-bin (--configure):
#17 43.24  installed libc-bin package post-installation script subprocess
returned error exit status 139

Exit 139 = SIGSEGV -- libc-bin's postinst script is segfaulting
mid-install, not a dependency or build-config problem. This looks like a
QEMU/emulation crash tied to that particular base-image/arch combination
rather than anything in our Dockerfile or Solr code; the job log also shows
binfmt/buildx cache contention around the same window. It's been
intermittent -- failing on several different matrix legs on different
nights over the past week or so, succeeding on others -- consistent with
flaky infra rather than a single introduced regression.

Flagging since it's currently the CI signal blocking backport tooling from
picking up any commit newer than the last green docker-nightly run. Anyone
with context on the GH Actions runner/QEMU side, input welcome.

note: ever since the docker nightly build was converted from a Jenkins Job
to a GHA runner, it no longer shows up in the [email protected] list

🤖 Generated with Claude Code (https://claude.com/claude-code)

~ David Smiley
Apache Lucene/Solr Search Developer
http://www.linkedin.com/in/davidwsmiley

Reply via email to