Build hdf5-parallel-openmpi.powerpc64le-linux on master is broken.

  • Done
  • quality assurance status badge
Details
2 participants
  • Ludovic Courtès
  • Maxim Cournoyer
Owner
unassigned
Submitted by
Maxim Cournoyer
Severity
normal
M
M
Maxim Cournoyer wrote on 8 Jan 2024 04:19
(name . bug-guix)(address . bug-guix@gnu.org)
87o7dwy0jw.fsf@gmail.com
Hello,

cuirass@gnu.org (Cuirass) writes:

Toggle quote (6 lines)
> <p>The build <b>hdf5-parallel-openmpi.powerpc64le-linux</b> for specification <b>master</b>
> is broken. You can find the detailed information about this build <a
> href="https://ci.guix.gnu.org/build/3175009/details">here</a>.</p>
>
> https://ci.guix.gnu.org/build/3175009/details

Sample output:

Toggle snippet (63 lines)
Atomicity Test Failed Process 4: read_buf[2038] is 5, should be 0
Atomicity Test Failed Process 4: read_buf[2039] is 5, should be 0
Atomicity Test Failed Process 4: read_buf[2040] is 5, should be 0
Atomicity Test Failed Process 4: read_buf[2041] is 5, should be 0
Atomicity Test Failed Process 4: read_buf[2042] is 5, should be 0
Atomicity Test Failed Process 4: read_buf[2043] is 5, should be 0
Atomicity Test Failed Process 4: read_buf[2044] is 5, should be 0
Atomicity Test Failed Process 4: read_buf[2045] is 5, should be 0
Atomicity Test Failed Process 4: read_buf[2046] is 5, should be 0
Atomicity Test Failed Process 4: read_buf[2047] is 5, should be 0
Testing -- Store Dense Attributes (denseattr)
Testing -- Store Dense Attributes (denseattr)
Testing -- Store Dense Attributes (denseattr)
Testing -- Store Dense Attributes (denseattr)
Testing -- Store Dense Attributes (denseattr)
Testing -- Store Dense Attributes (denseattr)
Testing -- Collective Metadata read with some ranks having no selection (noselcollmdread)
Testing -- Collective Metadata read with some ranks having no selection (noselcollmdread)
Testing -- Collective Metadata read with some ranks having no selection (noselcollmdread)
Testing -- Collective Metadata read with some ranks having no selection (noselcollmdread)
Testing -- Collective Metadata read with some ranks having no selection (noselcollmdread)
Testing -- Collective Metadata read with some ranks having no selection (noselcollmdread)
Testing -- Collective MD read with multi chunk I/O (H5D__chunk_addrmap) (MC_coll_MD_read)
Testing -- Collective MD read with multi chunk I/O (H5D__chunk_addrmap) (MC_coll_MD_read)
Testing -- Collective MD read with multi chunk I/O (H5D__chunk_addrmap) (MC_coll_MD_read)
Testing -- Collective MD read with multi chunk I/O (H5D__chunk_addrmap) (MC_coll_MD_read)
Testing -- Collective MD read with multi chunk I/O (H5D__chunk_addrmap) (MC_coll_MD_read)
Testing -- Collective MD read with multi chunk I/O (H5D__chunk_addrmap) (MC_coll_MD_read)
Testing -- Collective MD read with link chunk I/O (H5D__sort_chunk) (LC_coll_MD_read)
Testing -- Collective MD read with link chunk I/O (H5D__sort_chunk) (LC_coll_MD_read)
Testing -- Collective MD read with link chunk I/O (H5D__sort_chunk) (LC_coll_MD_read)
Testing -- Collective MD read with link chunk I/O (H5D__sort_chunk) (LC_coll_MD_read)
Testing -- Collective MD read with link chunk I/O (H5D__sort_chunk) (LC_coll_MD_read)
Testing -- Collective MD read with link chunk I/O (H5D__sort_chunk) (LC_coll_MD_read)


All tests were successful.



All tests were successful.



All tests were successful.



All tests were successful.



All tests were successful.



All tests were successful.

===================================
***PHDF5 tests detected 1040 errors***
===================================

--
Thanks,
Maxim
L
L
Ludovic Courtès wrote on 19 Aug 2024 12:37
Re: [bug#72071] [PATCH 0/2] A clean series of patches
(name . Luca Cirrottola)(address . luca.cirrottola@inria.fr)
871q2kx0go.fsf@gnu.org
Hi Luca,

Luca Cirrottola <luca.cirrottola@inria.fr> skribis:

Toggle quote (18 lines)
> Hello Ludo', following your suggestion, I have implemented a simpler
> solution that just skips off the problematic test (without moving
> around files). I send a new patch series with the full modifications
> to be applied to the recipe, containing one commit for the package
> version update, and one commit for skipping the test.
>
> Let me know if it looks better (no pressure).
>
> Thanks,
> Luca
>
> Luca Cirrottola (2):
> Solve issue https://issues.guix.gnu.org/68313 by updating HDF5 version
> to 1.14.3 (containing the official fix
> https://github.com/HDFGroup/hdf5/pull/3421).
> Skip test_check_version since the patch-settings phase needs to modify
> the test reference file.

I reorganized the changes a bit; in particular, since skipping the test
is necessary for 1.14.3 to build, I squashed the upgrade and the test
skip in the same commit.

Also, the test change would break hdf@1.8, so I moved that to hdf@1.14.

The result is:

84c62cf166 * gnu: hdf5-parallel-openmpi: Update to 1.14.x.
832f2bcb92 * gnu: hdf5@1.14: Update to 1.14.3.

Tested with:

./pre-inst-env guix build --no-grafts \
hdf5@1.8 hdf5@1.14 hdf5-parallel-openmpi hdf5@1.10

Thanks!

Ludo’.
Closed
?
Your comment

This issue is archived.

To comment on this conversation send an email to 68313@debbugs.gnu.org

To respond to this issue using the mumi CLI, first switch to it
mumi current 68313
Then, you may apply the latest patchset in this issue (with sign off)
mumi am -- -s
Or, compose a reply to this issue
mumi compose
Or, send patches to this issue
mumi send-email *.patch