DAOS-18609 test: Resolving slurm issue for el9 - #18294
Conversation
After CI moved over to el9 slurm package installation needed readjusting. slurm-example-configs removed slurm-slurmdbd added Quick-build: true Quick-Functional: true Test-tag: test_soak_smoke Signed-off-by: Saurabh Tandan <saurabh.tandan@hpe.com>
|
Ticket title is 'launch/functional_hardware_large - Error setting up slurm - Unable to find a match: slurm-example-configs' |
|
Test stage Functional Hardware Large MD on SSD completed with status FAILURE. https://jenkins-3.daos.hpc.amslabs.hpecorp.net//job/daos-stack/job/daos/view/change-requests/job/PR-18294/1/execution/node/742/log |
Quick-build: true Quick-Functional: true Test-tag: test_soak_smoke Signed-off-by: Saurabh Tandan <saurabh.tandan@hpe.com>
|
Test stage Functional Hardware Large MD on SSD completed with status FAILURE. https://jenkins-3.daos.hpc.amslabs.hpecorp.net//job/daos-stack/job/daos/view/change-requests/job/PR-18294/2/execution/node/742/log |
Quick-build: true Quick-Functional: true Test-tag: test_soak_smoke Signed-off-by: Saurabh Tandan <saurabh.tandan@hpe.com>
|
Test stage Functional Hardware Large MD on SSD completed with status FAILURE. https://jenkins-3.daos.hpc.amslabs.hpecorp.net//job/daos-stack/job/daos/view/change-requests/job/PR-18294/3/execution/node/742/log |
Quick-build: true Quick-Functional: true Test-tag: test_soak_smoke Signed-off-by: Saurabh Tandan <saurabh.tandan@hpe.com>
Quick-build: true Quick-Functional: true Test-tag: test_soak_smoke Signed-off-by: Saurabh Tandan <saurabh.tandan@hpe.com>
|
Test stage Functional Hardware Large MD on SSD completed with status FAILURE. https://jenkins-3.daos.hpc.amslabs.hpecorp.net//job/daos-stack/job/daos/view/change-requests/job/PR-18294/5/execution/node/779/log |
|
Test stage Test RPMs on EL 9.6 completed with status FAILURE. https://jenkins-3.daos.hpc.amslabs.hpecorp.net//job/daos-stack/job/daos/view/change-requests/job/PR-18294/6/execution/node/558/log |
Test-tag: test_soak_smoke
a17096f to
281778b
Compare
|
Test stage Functional Hardware Large MD on SSD completed with status FAILURE. https://jenkins-3.daos.hpc.amslabs.hpecorp.net//job/daos-stack/job/daos/view/change-requests/job/PR-18294/8/execution/node/1029/log |
is not being picked for el9 Test-tag: test_soak_smoke Signed-off-by: Saurabh Tandan <saurabh.tandan@hpe.com>
Test-tag: test_soak_smoke Signed-off-by: Saurabh Tandan <saurabh.tandan@hpe.com>
|
Test stage Functional Hardware Large MD on SSD completed with status FAILURE. https://jenkins-3.daos.hpc.amslabs.hpecorp.net//job/daos-stack/job/daos/view/change-requests/job/PR-18294/9/execution/node/1028/log |
phender
left a comment
There was a problem hiding this comment.
The majority of these changes don't seem to affect fixing the slurm RPM install issue:
2026/08/20 04:26:23 INFO install_packages: Installing packages on hdr-[140,142-149]: slurm, slurm-example-configs, slurm-slurmctld, slurm-slurmd
2026/08/20 04:26:23 DEBUG run_remote: Running on hdr-[140,142-149] with a 600 second timeout: sudo -n dnf install -y slurm slurm-example-configs slurm-slurmctld slurm-slurmd
2026/08/20 04:26:26 DEBUG log_result_data: hdr-144 (rc=1):
2026/08/20 04:26:26 DEBUG log_result_data: created by dnf config-manager from https://arti 82 kB/s | 3.0 kB 00:00
2026/08/20 04:26:26 DEBUG log_result_data: daos_ci-rocky9-base-artifactory 5.1 kB/s | 2.0 kB 00:00
2026/08/20 04:26:26 DEBUG log_result_data: daos_ci-rocky9-extras-artifactory 2.7 kB/s | 1.5 kB 00:00
2026/08/20 04:26:26 DEBUG log_result_data: daos_ci-rocky9-appstream-artifactory 15 kB/s | 2.4 kB 00:00
2026/08/20 04:26:26 DEBUG log_result_data: daos_ci-rocky9-crb-artifactory 7.6 kB/s | 2.4 kB 00:00
2026/08/20 04:26:26 DEBUG log_result_data: daos_ci-el9-epel-artifactory 93 kB/s | 4.0 kB 00:00
2026/08/20 04:26:26 DEBUG log_result_data: daos-stack-deps-el-9-stable-local-artifactory 5.2 kB/s | 478 B 00:00
2026/08/20 04:26:26 DEBUG log_result_data: daos_ci-intel-tools-el-8-stable-local-artifacto 5.1 kB/s | 478 B 00:00
2026/08/20 04:26:26 DEBUG log_result_data: created by dnf config-manager from https://jenk 77 kB/s | 3.0 kB 00:00
2026/08/20 04:26:26 DEBUG log_result_data: No match for argument: slurm-example-configs
2026/08/20 04:26:26 DEBUG log_result_data: Error: Unable to find a match: slurm-example-configs
Have you tried removing slurm-example-configs from https://github.com/daos-stack/daos/blob/master/src/tests/ftest/slurm_setup.py#L37
There was a problem hiding this comment.
The copyright needs updating - check your githooks
| (C) Copyright 2025-2026 Hewlett Packard Enterprise Development LP |
| return commands | ||
|
|
||
|
|
||
| def create_app_cmdline(self, job_spec, pool, ppn, nodesperjob): |
There was a problem hiding this comment.
We should add a `# pylint: disable=too-many-locals
|
|
||
| # Restart slurmd on all nodes | ||
| self._restart_systemctl(self.all_nodes, 'slurmd', '/var/log/slurmd.log', self.SLURM_CONF) | ||
| self._restart_systemctl( |
There was a problem hiding this comment.
Why does this need to be modified? The original line length does not exceed the limit.
| self._modify_slurm_config_file( | ||
| 'epilog file', self.all_nodes, 's#EpilogSlurmctld=#EpilogSlurmctld={EPILOG_FILE}#g', | ||
| self.root) | ||
| 'epilog file', self.all_nodes, |
| if "Socket" in info and "Core" in info and "Thread" in info: | ||
| echo_command = (f'echo \"Nodename={nodelist} Sockets={info["Socket"]} ' | ||
| f'CoresPerSocket={info["Core"]} ThreadsPerCore={info["Thread"]}\"') | ||
| f'CoresPerSocket={info["Core"]} ThreadsPerCore={info["Thread"]}\" ') |
There was a problem hiding this comment.
Is this extra space needed?
| os_name = (os_info.name or "").lower() | ||
| os_version = str(os_info.version or "").strip() | ||
|
|
||
| # Fallback for cases where avocado detect() returns unknown/0. |
There was a problem hiding this comment.
Where are we running into cases where avocado detect() returns unknown/0? We use detect() in every CI test in https://github.com/daos-stack/daos/blob/master/src/tests/ftest/util/apricot/apricot/test.py#L156 and it works with all of the distros we run on.
There was a problem hiding this comment.
Note: Discard results for Build#9 as wrong slurm_setup.py was accidentally pushed.
Regarding detect(), as noticed for the results of build#8, it was not able to detect the correct os version for el9 hence was not able to copy the correct lammps build hence lammps jobs failed. So I am just trying to bullet proof that in scenarios where it is failing to detect is especially in case of el9. Not sure what exactly is detect() used for in other tests, but here I need the version number as well.
| os_name = (os_release.get("NAME") or os_release.get("ID") or os_name).lower() | ||
| os_version = (os_release.get("VERSION_ID") or os_version).strip() | ||
|
|
||
| if os.environ.get("DAOS_TEST_MODE") is None: |
There was a problem hiding this comment.
This may be a bit out of scope for this PR, but it seems all DAOS_TEST_MODE does is prevent the code from updating DAOS_TEST_APP_DIR and I_MPI_OFI_LIBRARY_INTERNAL in certain situations. It seems like the updating of those variables should be dependent upon whether or not they are already set - and then we don't need DAOS_TEST_MODE.
Test-tag: test_soak_smoke Signed-off-by: Saurabh Tandan <saurabh.tandan@hpe.com>
Test-tag: test_soak_smoke Signed-off-by: Saurabh Tandan <saurabh.tandan@hpe.com>
Yeah I did in the previous commits. Along with removal of
Yeah I did, along with removal |
Test-tag: test_soak_smoke Signed-off-by: Saurabh Tandan <saurabh.tandan@hpe.com>
After CI moved over to el9 slurm package
installation needed readjusting.
slurm-example-configs removed
slurm-slurmdbd added
Quick-build: true
Quick-Functional: true
Test-tag: test_soak_smoke
Steps for the author:
After all prior steps are complete: