Description of problem: Every time I have to manually review TPS results and waive all TPS verify test failures on all architectures in order to be able to test a new kexec-tools pacakge in errata. Could you do me a favour to fix this trivial RPM spec file issue? The failure is, result: verify test: FAIL: kexec-tools.ppc64: /etc/kdump.conf S.5....T c [tps:B] The fix should be very simple. Configuration files that are created on the fly should be marked 'noverify' and thus not generate a failure. Example: %config(missingok,noreplace) %verify(not md5 size mtime) /etc/X11/xorg.conf The %ghost option may also be of use. Version-Release number of selected component (if applicable): kexec-tool-1.102pre-21.el5 How reproducible: always
I can, although this seems like more of an rpm issue, rather than a kexec-tools issue. /etc/kdump.conf isn't generated on the fly. Its part of the package, I just touch it in the %post script to update the mtime on the file so as to force an initramfs rebuild when the service restarts after an update (since we don't replace the config file in that case). Since its marked as a config file with the noreplace option, it seems like rpm shouldn't expect it to have the same mtime as whats in the updated package, nor should it have the same size or md5sum. I can mark it to not be verified, but this should likely be taken up with the rpm maintainer.
Corrected in -49.el5. Please look into open a bug on rpm for this however. Thanks
(In reply to comment #1) > I can, although this seems like more of an rpm issue, rather than a kexec-tools > issue. /etc/kdump.conf isn't generated on the fly. Its part of the package, I > just touch it in the %post script to update the mtime on the file so as to > force an initramfs rebuild when the service restarts after an update (since we > don't replace the config file in that case). Since its marked as a config file > with the noreplace option, it seems like rpm shouldn't expect it to have the > same mtime as whats in the updated package, nor should it have the same size or > md5sum. I can mark it to not be verified, but this should likely be taken up > with the rpm maintainer. Neil, I don't think %verify will check with %config first to decide what things to verify. You need to tell it explicitly what not to verify if you think it will give false alarm. From the document here, http://www.rpm.org/max-rpm/s1-rpm-inside-files-list-directives.html By default, it just verify everything. Sorry, but there is one more similar fix is needed for TPS to pass on IA-64, because /etc/sysconfig/kdump has also been modified there. I don't see it on other architectures though.
>Neil, I don't think %verify will check with %config first Thats really rather my point. I've flagged a file in kexec-tools as bineg a %config file, yet rpm isn't smart enough to realize that its verifications will likely fail on it? That seems like an RPM shortcomming to me. Regardless however As for the sysconfig/kdump touching, that only for HP zx1 machines, it will pass on all other ia64 systesm. But since apparently this is important enough to squeeze into 5.3 I'll go through the whole process again. Fixed in -50.el5
An advisory has been issued which should help the problem described in this bug report. This report is therefore being closed with a resolution of ERRATA. For more information on therefore solution and/or where to find the updated files, please follow the link below. You may reopen this bug report if the solution does not work for you. http://rhn.redhat.com/errata/RHBA-2009-0105.html