-
Bug
-
Resolution: Done
-
Major
-
CentOS Stream 10
-
None
-
None
-
Critical
-
rhel-pt-c-libs
-
ssg_platform_tools
-
1
-
False
-
False
-
-
None
-
None
-
Requested
-
None
-
-
All
-
None
What were you trying to do that didn't work?
While bootstrapping CentOS Stream 10, we've encountered a number of issues with package builds where there tests were failing which we tracked down to ld.so.cache being unavailable. This is due to the following error in the root.log:
DEBUG util.py:446: error: lua script failed: [string "%post(glibc-2.38.9000-22.eln132.x86_64)"]:13: attempt to call a nil value (field 'execute')
DEBUG util.py:444: Error in POSTIN scriptlet in rpm package glibc
DEBUG util.py:446: error: lua script failed: [string "%transfiletriggerin(glibc-common-2.38.9000-22.eln132.x86_64)"]:13: attempt to call a nil value (field 'execute')
DEBUG util.py:444: Error in <unknown> scriptlet in rpm package glibc-common
The root cause here is that about two years ago, this commit was merged, which changed the behavior of the scriptlets to use the rpm.execute() function to run ldconfig rather than the older pure-lua implementation. This worked fine on Fedora Koji where we have support for the use_bootstrap=True feature of mock, but we don't have proper support for that in CS Koji and RHEL Brew (and adding it is proving to introduce more issues than it solves). Since the CS and RHEL build instances only have RPM 4.14, this causes the scriptlets to fail.
This is becoming a blocker for the RHEL 10 bootstrapping efforts, so we would very much appreciate this being looked into immediately.
Please provide the package NVR for which bug is seen:
glibc-2.38.9000-26.el10+1
How reproducible:
Every time
Steps to reproduce
- Install glibc on CentOS Stream 10
Expected results
glibc is installed correctly
Actual results
See above error in %post and %transfiletriggerin