Bug 1085727

Summary: coreutils fails to build on XFS (AArch64 builders)
Product: [Fedora] Fedora Reporter: Marcin Juszkiewicz <mjuszkie>
Component: coreutilsAssignee: Ondrej Vasik <ovasik>
Status: CLOSED RAWHIDE QA Contact: Fedora Extras Quality Assurance <extras-qa>
Severity: unspecified Docs Contact:
Priority: unspecified    
Version: rawhideCC: admiller, kdudka, kzak, ooprala, ovasik, pbrady, p, twaugh
Target Milestone: ---   
Target Release: ---   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard:
Fixed In Version: coreutils-8.22-14.fc21 Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2014-04-13 03:41:31 UTC Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:
Attachments:
Description Flags
patch + spec changes
none
patch + spec
none
avoid XFS speculative preallocation test issues none

Description Marcin Juszkiewicz 2014-04-09 08:22:02 UTC
Created attachment 884433 [details]
patch + spec changes

Description of problem:

Coreutils builds fine on AArch64. But builders for this architecture are using XFS when rest of Fedora builders are Ext4.

This creates an issue cause dd/sparse.sh test fails on XFS (works on tmpfs, ext4). One of solutions is to disable test which this patch does.

Version-Release number of selected component (if applicable):

8.22-13

How reproducible:

always on xfs

Steps to Reproduce:
1. create xfs filesystem
2. build coreutils package
3. watch it fails

Actual results:

FAIL: tests/dd/sparse.sh

Expected results:

package builds

Additional info:

Comment 1 Ondrej Vasik 2014-04-09 08:51:08 UTC
Thanks for report, I believe I have already seen such report somewhere - can't find it now, though. No need for attaching such patches - especially when you attach just the spec file changes and not the patch.
I think disabling tests for builders is perfectly fine temporary solution - and we already disable some tests due to various koji build failures, but maybe we can find something more generic here.

Comment 2 Ondrej Vasik 2014-04-09 08:54:17 UTC
Ah, now I see... original one is https://bugzilla.redhat.com/show_bug.cgi?id=1075679 , RHEL-6 - with several links to upstream commits supposed to fix that. If the issue is still not fixed, probably fixes need to be extended to cover even the current failures.

Comment 3 Marcin Juszkiewicz 2014-04-09 09:15:25 UTC
Created attachment 884442 [details]
patch + spec

oops, forgot to add patch part.

Patches mentioned in bug 1075679 are part of 8.22 release.

Comment 4 Ondrej Vasik 2014-04-11 15:15:32 UTC
Yes, I know they are part of 8.22 - I just wanted to point out that upstream likely intends to fix the test for xfs/nfs and skip only necessary parts - not whole test based on the fstype. We will see - for now I'm fine with just skipping the test in koji builds.

Comment 5 Pádraig Brady 2014-04-11 17:15:09 UTC
This is indicating v surprising behavior on XFS.
I've sent a query to the xfs mailing list.
Will workaround upstream in any case.

Comment 6 Pádraig Brady 2014-04-11 23:02:36 UTC
Created attachment 885638 [details]
avoid XFS speculative preallocation test issues

Details of this discussion at thread starting at:
http://oss.sgi.com/archives/xfs/2014-04/msg00285.html

Comment 7 Ondrej Vasik 2014-04-13 03:41:31 UTC
Committed to distgit -> https://lists.fedoraproject.org/pipermail/scm-commits/Week-of-Mon-20140407/1223907.html -> CLOSED RAWHIDE. Thanks Pádraig!