On Wed, 20 Feb 2019 15:08:24 -0500
Kaleb Keithley <kkeithle(a)redhat.com> wrote:
I'm trying to build Ceph again for f31 after the branching.
It built before the branching, eight days ago.
The x86_64[1] part of number two below got further than either of the
two examples below before I killed it. I'm guessing it would have
finished successfully if I had let it. I have another running now to
confirm.
But I'm getting strange errors on s390x.
One, e.g. (
https://kojipkgs.fedoraproject.org//work/tasks/6799/32926799/build.log)
BUILDSTDERR: {standard input}: Assembler messages:
BUILDSTDERR: {standard input}:464309: Warning: end of file not at end
of a line; newline inserted
BUILDSTDERR: {standard input}:465519: Error: invalid operands (*UND*
and .gcc_except_table sections) for `-'
BUILDSTDERR: c++: fatal error: Killed signal terminated program
cc1plus BUILDSTDERR: compilation terminated.
An earlier one
(
https://kojipkgs.fedoraproject.org//work/tasks/3857/32923857/build.log)
BUILDSTDERR: {standard input}: Error: .size expression for
_ZNSt8_Rb_treeINSt7__cxx1112basic_stringIcSt11char_traitsIcESaIcEEESt4pairIKS5_16StringConstraintESt10_Select1stIS9_ESt4lessIS5_ESaIS9_EE7_M_copyINSF_20_Reuse_or_alloc_nodeEEEPSt13_Rb_tree_nodeIS9_EPKSJ_PSt18_Rb_tree_node_baseRT_
does not evaluate to a constant
BUILDSTDERR: c++: fatal error: Killed signal terminated program
cc1plus BUILDSTDERR: compilation terminated.
Bad disk? Out of disk space? Any thoughts?
this usually means out of memory when compiling big C++ projects, c++
is killed by the kernel. So either you need to make the build
less-parallel, you can reduce the size of the debuginfo (the s/-g/-g1/
trick) or there need to be more memory added to the builders.
Please post the link to the build task, not to the logs, next time.
Dan