strange koji build failures...?

Alex Lancaster alexl at users.sourceforge.net
Wed Jun 4 09:01:50 UTC 2008


>>>>> "OV" == Ondřej Vašík  writes:

OV> Alex Lancaster wrote:
MT> Maybe coreutils-6.12-1.fc10 should be untagged?
>> > Perhaps the coreutils on Koji's mock pulls from somewhere else...
>> I can also confirm two similar build failures, both with "touch":
>> 
>> f-spot:
>> http://koji.fedoraproject.org/koji/getfile?taskID=644642&name=build.log

OV> Problem likely caused by gl_futimens() (which is used in
OV> install,cp,mv and touch) and quiet old koji xen kernel.  I had
OV> issues with failures of touch test suite in the past - as
OV> futimensat call returned absolute nonsense number(filled as
OV> #442352). It seems to be changed - now kernel returns ENOSYS(at
OV> least it seemed so as coreutils-6.12 test suite was failing due
OV> this). I used recent patch from upstream which fixed test suite
OV> failures. Now it seems that the issue is not fully fixed and it
OV> still needs some improvement. Strace of the failure is on the
OV> way(building ;) ) and I hope it will help me to at least
OV> workaround the issue with koji build asap.

For others following this, I filed a bug against the actual coreutils
component:

https://bugzilla.redhat.com/show_bug.cgi?id=449910

Alex




More information about the devel mailing list