gnu-compilers: test build with tmate debug on failure

[adrianreber]

Add tmate SSH debug session to the openEuler build job to investigate brp-strip failure on aarch64. Includes a dummy spec change to trigger the gnu-compilers build.

[github-actions]

📦 Package Count Analysis Results

Environment: UBI 10 Container
Target Branch: 4.x
Version: OpenHPC 4.x
Overall Status: ✅ Analysis completed successfully

🏭 Factory Repositories

Status: ✅ Factory repositories analysis completed successfully

Release	Base OS	aarch64	x86_64	noarch	Total
4.2	EL_10	54	95	12	161
4.2	openEuler_24.03	48	55	12	115
Total	All	102	150	24	276

---

Analysis performed by OpenHPC Package Count CI
Commit: 75cedff5
Workflow: 28771689832

[github-actions]

📊 Build Order Analysis Results

Environment: AlmaLinux 10 Container
Total Spec Files Analyzed: 89
Status: ✅ Build order analysis completed successfully

📋 Build Order Output

📦 Build Order (84 packages)

• 📄 ohpc-filesystem.spec
• 📄 lmod.spec
• 📄 nhc.spec
• 📄 prun.spec
• 📄 mrsh.spec
• 📄 losf.spec
• 📄 examples.spec
• 📄 magpie.spec
• 📄 intel-compilers-devel.spec
• 📄 gnu-compilers.spec
• 📄 warewulf.spec
• 📄 cubew.spec
• 📄 flex.spec
• 📄 sigar.spec
• 📄 python-rpm-macros.spec

➕ Show remaining 69 packages

• 📄 python-Cython.spec
• 📄 papi.spec
• 📄 likwid.spec
• 📄 pdtoolkit.spec
• 📄 paraver.spec
• 📄 lustre.spec
• 📄 ucx.spec
• 📄 mvapich2.spec
• 📄 intel-mpi.spec
• 📄 mpich.spec
• 📄 cmake.spec
• 📄 hwloc.spec
• 📄 valgrind.spec
• 📄 cuda-devel.spec
• 📄 spack.spec
• 📄 easybuild.spec
• 📄 charliecloud.spec
• 📄 conman.spec
• 📄 tests.spec
• 📄 docs.spec
• 📄 genders.spec
• 📄 hpcworkspace.spec
• 📄 pmix.spec
• 📄 slurm.spec
• 📄 openblas.spec
• 📄 example.spec
• 📄 metis.spec
• 📄 cubelib.spec
• 📄 superlu.spec
• 📄 plasma.spec
• 📄 gsl.spec
• 📄 opari2.spec
• 📄 gotcha.spec
• 📄 msr-safe.spec
• 📄 python-numpy.spec
• 📄 pdsh.spec
• 📄 openpbs.spec
• 📄 R.spec
• 📄 openmpi.spec
• 📄 python-mpi4py.spec
• 📄 lmod-defaults.spec
• 📄 scotch.spec
• 📄 scalapack.spec
• 📄 mumps.spec
• 📄 opencoarrays.spec
• 📄 hypre.spec
• 📄 superlu_dist.spec
• 📄 fftw.spec
• 📄 boost.spec
• 📄 pnetcdf.spec
• 📄 sionlib.spec
• 📄 hdf5.spec
• 📄 netcdf.spec
• 📄 imb.spec
• 📄 extrae.spec
• 📄 dimemas.spec
• 📄 omb.spec
• 📄 python-scipy.spec
• 📄 petsc.spec
• 📄 trilinos.spec
• 📄 mfem.spec
• 📄 otf2.spec
• 📄 netcdf-cxx4.spec
• 📄 netcdf-fortran.spec
• 📄 adios2.spec
• 📄 tau.spec
• 📄 scorep.spec
• 📄 slepc.spec
• 📄 scalasca.spec

---

Analysis performed by OpenHPC Build Order CI
Commit: 75cedff5
Workflow: 28771689823

[github-actions]

🚀 CCache Statistics

Configuration	🐧 RHEL	🦊 openEuler
🖥️ x86_64 / gnu15	🟡 25.61%	🟡 24.81%
🖥️ x86_64 / intel	🟡 26.71%	-
💪 aarch64 / gnu15	🔴 16.69%	🔴 15.98%

📊 Detailed Statistics

openEuler (aarch64)

Cacheable calls:    5100 / 9244 (55.17%)
  Hits:              815 / 5100 (15.98%)
    Direct:          399 /  815 (48.96%)
    Preprocessed:    416 /  815 (51.04%)
  Misses:           4285 / 5100 (84.02%)
Uncacheable calls:  4144 / 9244 (44.83%)
Local storage:
  Cache size (GiB):  0.4 /  5.0 ( 7.52%)
  Hits:              815 / 5100 (15.98%)
  Misses:           4285 / 5100 (84.02%)

openEuler (x86_64)

Cacheable calls:    6199 / 11448 (54.15%)
  Hits:             1538 /  6199 (24.81%)
    Direct:          706 /  1538 (45.90%)
    Preprocessed:    832 /  1538 (54.10%)
  Misses:           4661 /  6199 (75.19%)
Uncacheable calls:  5249 / 11448 (45.85%)
Local storage:
  Cache size (GiB):  0.4 /   5.0 ( 8.18%)
  Hits:             1538 /  6199 (24.81%)
  Misses:           4661 /  6199 (75.19%)

RHEL (aarch64/gnu15)

Cacheable calls:    5153 / 9334 (55.21%)
  Hits:              860 / 5153 (16.69%)
    Direct:          441 /  860 (51.28%)
    Preprocessed:    419 /  860 (48.72%)
  Misses:           4293 / 5153 (83.31%)
Uncacheable calls:  4181 / 9334 (44.79%)
Local storage:
  Cache size (GiB):  0.4 /  5.0 ( 7.54%)
  Hits:              860 / 5153 (16.69%)
  Misses:           4293 / 5153 (83.31%)

RHEL (x86_64/gnu15)

Cacheable calls:    6270 / 11566 (54.21%)
  Hits:             1606 /  6270 (25.61%)
    Direct:          769 /  1606 (47.88%)
    Preprocessed:    837 /  1606 (52.12%)
  Misses:           4664 /  6270 (74.39%)
Uncacheable calls:  5296 / 11566 (45.79%)
Local storage:
  Cache size (GiB):  0.4 /   5.0 ( 8.20%)
  Hits:             1606 /  6270 (25.61%)
  Misses:           4664 /  6270 (74.39%)

RHEL (x86_64/intel)

Cacheable calls:     6422 / 16986 (37.81%)
  Hits:              1715 /  6422 (26.71%)
    Direct:           765 /  1715 (44.61%)
    Preprocessed:     950 /  1715 (55.39%)
  Misses:            4707 /  6422 (73.29%)
Uncacheable calls:  10564 / 16986 (62.19%)
Local storage:
  Cache size (GiB):   0.4 /   5.0 ( 7.02%)
  Hits:              1715 /  6422 (26.71%)
  Misses:            4707 /  6422 (73.29%)

---

🤖 Generated from workflow run 28771689862

[github-actions]

Test Results

 29 files   -  75  29 suites   - 75   53s ⏱️ - 5m 19s
 59 tests  - 242  53 ✅  - 242  6 💤 ±0  0 ❌ ±0 
105 runs   - 395  97 ✅  - 395  8 💤 ±0  0 ❌ ±0 

Results for commit 7cf6e88. ± Comparison against base commit b6af457.

This pull request removes 242 tests.

rm_execution ‑ [Boost/Accumulators] max-test under resource manager (slurm/gnu15/mpich)
rm_execution ‑ [Boost/Accumulators] max-test under resource manager (slurm/gnu15/mvapich2)
rm_execution ‑ [Boost/Accumulators] max-test under resource manager (slurm/gnu15/openmpi5)
rm_execution ‑ [Boost/Accumulators] max-test under resource manager (slurm/intel/mpich)
rm_execution ‑ [Boost/Accumulators] max-test under resource manager (slurm/intel/mvapich2)
rm_execution ‑ [Boost/Accumulators] max-test under resource manager (slurm/intel/openmpi5)
rm_execution ‑ [Boost/Accumulators] min-test under resource manager (slurm/gnu15/mpich)
rm_execution ‑ [Boost/Accumulators] min-test under resource manager (slurm/gnu15/mvapich2)
rm_execution ‑ [Boost/Accumulators] min-test under resource manager (slurm/gnu15/openmpi5)
rm_execution ‑ [Boost/Accumulators] min-test under resource manager (slurm/intel/mpich)
…

[viniciusferrao]

@adrianreber I was looking at this and it feels like a race condition in the parallel %__os_install_post. The binaries that fail are all hardlinks, so I think find-debuginfo / brp-strip ends up stripping both names of the same inode at once and corrupts it, which is why it's unreliable and only happens on the aarch64 runner. Have you tried it with -j1?

[adrianreber]

Closing this as it was just as test.

Have you tried it with -j1?

No, I haven't.