- Dec 09, 2020
-
-
Nate Rini authored
Match filtering requirement of sacctmgr show associations. Bug 10329
-
Nate Rini authored
Match filtering requirement of sacctmgr show associations. Bug 10329
-
Tim McMullan authored
This fixes an issue detecting the Mellanox provided UCX versions. Bug 10288.
-
Tim McMullan authored
Also don't require the RPM for such builds, since one likely doesn't exist. Bug 10288.
-
Tim Wickberg authored
Repeated calls to slurm_conf_init() (or slurm_init()) will not change the configuration. Only slurm_conf_reinit() is capable of such a change.
-
Tim Wickberg authored
-
Tim Wickberg authored
Add to test7.3 where it was missing so the test works again.
-
Tim Wickberg authored
This new slurm_init() function MUST be called before any internal API calls to ensure Slurm's internal configuration structures have been populated. (Internally this only calls slurm_conf_init() at present, but may handle additional initialization at some later date.) This replaces the prior ELF init mechanism which did not work correctly in a variety of use cases. This is required in 20.11 as the various slurm_get_*() calls have been removed, and these were indirectly triggering a call into slurm_conf_init() for most - but not all - API calls.
-
Tim McMullan authored
-
Tim Wickberg authored
This reverts commit 8ace44c2.
-
Tim Wickberg authored
This reverts commit 9487a942.
-
Tim Wickberg authored
This reverts commit ebdb10e7.
-
Tim Wickberg authored
This reverts commit a5a76857.
-
Tim Wickberg authored
This reverts commit 341b8bd0.
-
Tim Wickberg authored
This reverts commit d7fd18bb.
-
Nate Rini authored
Bug 10367
-
- Dec 08, 2020
-
-
Albert Gil authored
-
Scott Jackson authored
Bug 9007
-
Scott Jackson authored
Due to former use of exit_code in globals and with command timeouts Bug 9007
-
Scott Jackson authored
Because we no longer support using exit_code in globals. Refactored tests that were using fail_on_error Bug 9007
-
Scott Jackson authored
Refactored tests relying on global exit_code (from globals_accounting) Bug 9007
-
Scott Jackson authored
Replaced some routines with calls to get_config_param including: get_control_machine, slurmctld_plug_stack_nonstop, job_submit_all_partitions Removed unused proc copy_conf. Refactored tests formerly relying on setting of exit_code by functions in globals. Bug 9007
-
Scott Jackson authored
Bug 9007
-
Dominik Bartkiewicz authored
_foreach_reservation_license_list() skip reservation without licenses. Bug 10358
-
Dominik Bartkiewicz authored
Regression from f01f549b Bug 10235
-
Brian Christiansen authored
-
Brian Christiansen authored
Account for adjustments to allow "b" after multipliers. Bug 10330
-
- Dec 07, 2020
-
-
Marcin Stolarek authored
Support 'b' and 'B' as a suffix to memory specification. While it's not required we may have users submitting --mem=10GB and it may be difficult to understand why that stopped working. Bug 10330
-
- Dec 05, 2020
-
-
Nate Rini authored
When a connection reconnects, it can revert to the default sql_mode losing the explicit settings done at connection startup. When commit 0c5ead06 fixed the reconnect setting code, it inadvertently caused a potential race condition where sql_mode settings can get lost mid-connection. Bug 10342
-
- Dec 04, 2020
-
-
Tim Wickberg authored
-
Tim Wickberg authored
If not set, the plugstack option will be populated from get_extra_conf_path() instead, which means these direct tests on the option will fail and the SPANK prolog/epilog will not run. Regression from abd4db77 where the default option was no longer always filled in. Bug 9081. Signed-off-by:
Tim McMullan <mcmullan@schedmd.com>
-
Tim Wickberg authored
-
Brian Christiansen authored
-
Marcin Stolarek authored
Before that the test was actually only testing if srun --mail-type=all is accepted, after those changes we make a basic verification if MailProg was called with appropriate arguments for begin and end mails. Bug 9988
-
Marcin Stolarek authored
The sleep before seff call is to let the job go to COMPLETED (not COMPLETING) state and store the information in slurmdbd. 5 seconds should be enough on healthy installation. The sleep is async from slurmctld perspective, but it's in the systemd created cgroup (the slice) of slurmctld and operations like `systemctl restart slurmctld` will wait till its completition. Bug 9988
-
Nate Rini authored
Bug 10050
-
- Dec 03, 2020