Hello,
I've been trying to debug a segfault during %check that only occurs in epel9
Koji, but not in mock.
At the end, I compared the list of packages with:
$ koji list-buildroot 34845388 | sort > koji
$ mock -r centos-stream+epel-9-x86_64 shell -- rpm -qa | sort > mock
$ diff -u koji mock | grep -v ' '
+acl-2.3.1-3.el9.x86_64
-audit-libs-3.0.7-101.el9.x86_64
+audit-libs-3.0.7-102.el9.x86_64
-binutils-gold-2.35.2-17.el9.x86_64
-binutils-2.35.2-17.el9.x86_64
+binutils-gold-2.35.2-19.el9.x86_64
+binutils-2.35.2-19.el9.x86_64
-centos-gpg-keys-9.0-9.el9.noarch
-centos-stream-release-9.0-9.el9.noarch
-centos-stream-repos-9.0-9.el9.noarch
+centos-gpg-keys-9.0-12.el9.noarch
+centos-stream-release-9.0-12.el9.noarch
+centos-stream-repos-9.0-12.el9.noarch
-crypto-policies-20220203-1.gitf03e75e.el9.noarch
+crypto-policies-20220404-1.git845c0c1.el9.noarch
+cryptsetup-libs-2.4.3-4.el9.x86_64
-cyrus-sasl-lib-2.1.27-19.el9.x86_64
+cyrus-sasl-lib-2.1.27-20.el9.x86_64
+dbus-broker-28-5.el9.x86_64
+dbus-common-1.12.20-5.el9.noarch
+dbus-1.12.20-5.el9.x86_64
+device-mapper-libs-1.02.183-4.el9.x86_64
+device-mapper-1.02.183-4.el9.x86_64
-elfutils-debuginfod-client-0.186-1.el9.x86_64
-elfutils-default-yama-scope-0.186-1.el9.noarch
-elfutils-libelf-0.186-1.el9.x86_64
-elfutils-libs-0.186-1.el9.x86_64
-elfutils-0.186-1.el9.x86_64
-epel-release-9-2.el9.noarch
+elfutils-debuginfod-client-0.186-3.el9.x86_64
+elfutils-default-yama-scope-0.186-3.el9.noarch
+elfutils-libelf-0.186-3.el9.x86_64
+elfutils-libs-0.186-3.el9.x86_64
+elfutils-0.186-3.el9.x86_64
-expat-2.2.10-9.el9.x86_64
-fedpkg-minimal-1.2.0-4.el9.noarch
+expat-2.2.10-10.el9.x86_64
-flexiblas-netlib-3.0.4-7.el9.x86_64
-flexiblas-openblas-openmp-3.0.4-7.el9.x86_64
-flexiblas-3.0.4-7.el9.x86_64
+flexiblas-netlib-3.0.4-8.el9.x86_64
+flexiblas-openblas-openmp-3.0.4-8.el9.x86_64
+flexiblas-3.0.4-8.el9.x86_64
-glibc-common-2.34-25.el9.x86_64
-glibc-gconv-extra-2.34-25.el9.x86_64
-glibc-minimal-langpack-2.34-25.el9.x86_64
-glibc-2.34-25.el9.x86_64
+glibc-common-2.34-29.el9.x86_64
+glibc-gconv-extra-2.34-29.el9.x86_64
+glibc-minimal-langpack-2.34-29.el9.x86_64
+glibc-2.34-29.el9.x86_64
-gnutls-3.7.3-6.el9.x86_64
+gnutls-3.7.3-9.el9.x86_64
+gpg-pubkey-3228467c-613798eb
+gpg-pubkey-8483c65d-5ccc5b19
-krb5-libs-1.19.1-13.el9.x86_64
+kmod-libs-28-7.el9.x86_64
+krb5-libs-1.19.1-15.el9.x86_64
-libgcc-11.2.1-9.4.el9.x86_64
-libgcrypt-1.10.0-2.el9.x86_64
-libgfortran-11.2.1-9.4.el9.x86_64
-libgomp-11.2.1-9.4.el9.x86_64
+libgcc-11.2.1-10.el9.x86_64
+libgcrypt-1.10.0-3.el9.x86_64
+libgfortran-11.2.1-10.el9.x86_64
+libgomp-11.2.1-10.el9.x86_64
-libquadmath-11.2.1-9.4.el9.x86_64
+libquadmath-11.2.1-10.el9.x86_64
+libseccomp-2.5.2-2.el9.x86_64
-libstdc++-11.2.1-9.4.el9.x86_64
+libstdc++-11.2.1-10.el9.x86_64
-libxml2-2.9.12-4.el9.x86_64
+libxml2-2.9.13-1.el9.x86_64
-openblas-openmp-0.3.15-2.el9.x86_64
+openblas-openmp-0.3.15-3.el9.x86_64
-openblas-0.3.15-2.el9.x86_64
-openldap-2.4.59-3.el9.x86_64
-openssl-libs-3.0.1-12.el9.x86_64
-openssl-3.0.1-12.el9.x86_64
+openblas-0.3.15-3.el9.x86_64
+openldap-2.4.59-4.el9.x86_64
+openssl-libs-3.0.1-18.el9.x86_64
+openssl-3.0.1-18.el9.x86_64
-pyproject-rpm-macros-1.0.0~rc1-1.el9.noarch
+pyproject-rpm-macros-1.0.1-1.el9.noarch
-systemd-libs-250-3.el9.x86_64
+systemd-libs-250-4.el9.x86_64
+systemd-pam-250-4.el9.x86_64
+systemd-rpm-macros-250-4.el9.noarch
+systemd-250-4.el9.x86_64
-tpm2-tss-3.0.3-7.el9.x86_64
-zlib-1.2.11-31.el9.x86_64
+zlib-1.2.11-32.el9.x86_64
This seems like my local mock has newer c9s packages than the Koji build repo.
Is that expected, or is pulling c9s packages into the build repo stuck on Koji
side?
Actually, I got an idea that EPEL 9 Koji might already be using (internal?)
RHEL 9.0, possibly I have missed this switch... However, the
centos-stream-release package contraditcs taht idea :/
I've checked with an EPEL 9 Next Koji scratchbuild and it got e.g.
pyproject-rpm-macros-1.0.1-1.el9.
--
Miro Hrončok
--
Phone: +420777974800
IRC: mhroncok