DAOS-15596: Update to argobots 1.2

[knard-intel]

Description

• Update to argobots 1.2
• Update packaging/
• Add patch 411e5b3 fixing DAOS-14248
• Add patch bb0c908 fixing libunwind support
• remove deprecated patch 7fd1987

[knard-intel]

The test failure with master is a known issue:

• test_daos_degraded_ec: Issue DAOS-16337 which is duplicate of DAOS-16336

Issue of the release 2.6 are also related to the same tickets: timeout not respected for an not yet defined reason.

[brianjmurrell]

Just one small change to revert the Makefile changes I think.

[brianjmurrell]

@knard-intel Please add @daos-stack/release-engineering as reviewer when you are ready for this to land.

[knard-intel]

@brianjmurrell , just discovered this morning a potential regression of the configuration option with argobots v1.2.
Going to check it.
To be more accurate, it seems that the option --enable-stack-unwind is not well managed with this new version.
Indeed, from the build log, the following warning is output:

+ /usr/bin/sed -i.backup -e 's~compiler_flags=$~compiler_flags="-Wl,-z,now -specs=/usr/lib/rpm/redhat/redhat-hardened-ld"~' ./m4/ltmain.sh
+ ./configure --build=x86_64-redhat-linux-gnu --host=x86_64-redhat-linux-gnu --program-prefix= --disable-dependency-tracking --prefix=/usr --exec-prefix=/usr --bindir=/usr/bin --sbindir=/usr/sbin --sysconfdir=/etc --datadir=/usr/share --includedir=/usr/include --libdir=/usr/lib64 --libexecdir=/usr/libexec --localstatedir=/var --sharedstatedir=/var/lib --mandir=/usr/share/man --infodir=/usr/share/info --enable-valgrind --enable-stack-unwind
configure: WARNING: unrecognized options: --enable-stack-unwind

I am able to reproduce it locally.
Note: I found them on working on a different task: integration of the asan lib.

[brianjmurrell]

@brianjmurrell , just discovered this morning a potential regression of the configuration option with argobots v1.2. Going to check it.

Sounds good. If there is a regression, it would be prudent for us to understand how it was not found in testing and rectify that.

[brianjmurrell]

-1 pending @knard-intel's investigation of a potential regression.

[knard-intel]

This PR is blocked until the PR pmodels/argobots#401 has been landed.

[knard-intel]

@brianjmurrell, the regression is not easy do identify: it does not affect the general behavior of argbots. From my understanding the unwind lib is a debugging functionality allowing to help for the debugging of argobots crash. The only trace of the regression is a warning displayed during the config step of the argobots building process. At this time, I am still trying to land the fix (which has already been approved).

We should be using --enable-option-checking=fatal in our configure arguments to detect and fail on this kind of thing. Could you temporarily revert your Patch2: and enable this option in argobots.spec and confirm that it would have found this regression?

Yep, I am going to do it. However, I will not be able to look at the result as I have still not received my new laptop.

Done with commit dbcdfee and cddd5b2

[brianjmurrell]

Done with commit dbcdfee and cddd5b2

It works!

[2024-09-17T07:02:38.206Z] + ./configure --build=x86_64-redhat-linux-gnu --host=x86_64-redhat-linux-gnu --program-prefix= --disable-dependency-tracking --prefix=/usr --exec-prefix=/usr --bindir=/usr/bin --sbindir=/usr/sbin --sysconfdir=/etc --datadir=/usr/share --includedir=/usr/include --libdir=/usr/lib64 --libexecdir=/usr/libexec --localstatedir=/var --sharedstatedir=/var/lib --mandir=/usr/share/man --infodir=/usr/share/info --enable-valgrind --enable-stack-unwind --enable-option-checking=fatal
[2024-09-17T07:02:38.206Z] configure: error: unrecognized options: --enable-stack-unwind
[2024-09-17T07:02:38.206Z] error: Bad exit status from /var/tmp/rpm-tmp.o8heuC (%build)
[2024-09-17T07:02:38.206Z]     Bad exit status from /var/tmp/rpm-tmp.o8heuC (%build)
[2024-09-17T07:02:38.206Z] RPM build errors:
[2024-09-17T07:02:38.206Z] Child return code was: 1

Great job @knard-intel.

[knard-intel]

Done with commit dbcdfee and cddd5b2

It works!

[2024-09-17T07:02:38.206Z] + ./configure --build=x86_64-redhat-linux-gnu --host=x86_64-redhat-linux-gnu --program-prefix= --disable-dependency-tracking --prefix=/usr --exec-prefix=/usr --bindir=/usr/bin --sbindir=/usr/sbin --sysconfdir=/etc --datadir=/usr/share --includedir=/usr/include --libdir=/usr/lib64 --libexecdir=/usr/libexec --localstatedir=/var --sharedstatedir=/var/lib --mandir=/usr/share/man --infodir=/usr/share/info --enable-valgrind --enable-stack-unwind --enable-option-checking=fatal
[2024-09-17T07:02:38.206Z] configure: error: unrecognized options: --enable-stack-unwind
[2024-09-17T07:02:38.206Z] error: Bad exit status from /var/tmp/rpm-tmp.o8heuC (%build)
[2024-09-17T07:02:38.206Z]     Bad exit status from /var/tmp/rpm-tmp.o8heuC (%build)
[2024-09-17T07:02:38.206Z] RPM build errors:
[2024-09-17T07:02:38.206Z] Child return code was: 1

Great job @knard-intel.

OK, so lets try it without the faulty commit.
🚀

[knard-intel]

I am not able to investigate why the CI is failing.
I will do the investigation as soon as I have received my new company laptop.

[knard-intel]

I am able again to access the CI.
Restart it as the logs seem to not be available any more.

[knard-intel]

The only failure is a known issue with argobots stack overflow and UCX:

• All errors from https://build.hpdd.intel.com/blue/organizations/jenkins/daos-stack%2Fdaos/detail/ci-daos-stack-argobots-PR-26-master/9/tests/ are related to the same ticket https://daosio.atlassian.net/browse/DAOS-16608

[knard-intel]

Please @daos-stack/release-engineering could you lend this PR.
From my understanding, the PR daos-stack/daos#15181 should be landed asap after this one has been landed

[brianjmurrell]

Since this PR fixes more than one ticket, please also add Fixes: DAOS-14248 to your commit message so that the commit message makes it clear which additional issues are being fixed (i.e. in addition to the one mentioned in the commit subject).

[brianjmurrell]

Could you please update the datestamp here and add a comment about the 2nd patch that has been added and it's ticket number.

Please also feel free to change the author here to yourself since you did the most significant amount of work.

[knard-intel]

• Add reference to the new patch bb0c908 fixing configuration issue
• Update author and timestamp

[knard-intel]

• Add reference to the new patch bb0c908 fixing configuration issue
• Update author and timestamp

Fixed with commit b740042

[knard-intel]

I have restarted the stage which was failing on getting access to external resources.
The second run have succeed :-)

• https://build.hpdd.intel.com/blue/organizations/jenkins/daos-stack%2Fargobots/detail/PR-26/13/pipeline/
• https://build.hpdd.intel.com/blue/organizations/jenkins/daos-stack%2Fdaos/detail/ci-daos-stack-argobots-PR-26-release-2.6/2/pipeline/

Please @daos-stack/release-engineering could you please land this PR and its associated DAOS one daos-stack/daos#15181

[knard-intel]

Please @daos-stack/release-engineering could you please land this PR and its associated DAOS one daos-stack/daos#15181

[knard-intel]

All tests are passing of the latest build https://build.hpdd.intel.com/blue/organizations/jenkins/daos-stack%2Fargobots/detail/PR-26/14/pipeline/463 are passing. I have successfully restarted the failing test https://build.hpdd.intel.com/blue/organizations/jenkins/daos-stack%2Fdaos/detail/ci-daos-stack-argobots-PR-26-release-2.6/4/pipeline and it is passing with the build https://build.hpdd.intel.com/blue/organizations/jenkins/daos-stack%2Fdaos/detail/ci-daos-stack-argobots-PR-26-release-2.6/6/pipeline/

[brianjmurrell]

Please @daos-stack/release-engineering could you please land this PR

I will go ahead and do that. In the future, you can signal you are ready for landing by simply adding @daos-stack/release-engineering as a reviewer, similar to how you add @daos-stack/daos-gatekeeper to daos requests.

and its associated DAOS one daos-stack/daos#15181

I cannot land that one. You will need to request @daos-stack/daos-gatekeeper for that one. But I have also requested changes on that one so it will need another push before it's ready.