Fix the light tests failure on MacOS by lin-yang-ly · Pull Request #382 · idaholab/tmap8

lin-yang-ly · 2026-03-11T00:53:23Z

(Ref. #380)

moosebuild · 2026-03-11T01:10:45Z

Job Documentation, step Sync to remote on 0f0c001 wanted to post the following:

View the site here

This comment will be updated on new commits.

moosebuild · 2026-03-11T01:14:29Z

Job Coverage, step Generate coverage on ca367b7 wanted to post the following:

Coverage

Coverage did not change

Full coverage report

This comment will be updated on new commits.

lin-yang-ly · 2026-03-19T17:01:48Z

@simopier

Current PR attempt to solve two issues:

Test failures between Mac and HPC environments for val-2c, val-2f, and yttrium_hydrogen_system
Test failures between single-core and multi-core runs for val-2c and yttrium_hydrogen_system

The first issue is addressed by setting a reasonable rel_err (for val-2c), adjusting the trap_per_free parameter (for val-2f), and adding the automatic_scaling parameter (for yttrium_hydrogen_system).

The second issue is resolved by using PETSc solver parameters (-pc_type lu -pc_factor_mat_solver_type mumps).

For val-2f, the second issue is temporarily deferred, as the new PETSc solver parameters do not fully resolve it. Specifically, the val-2f_light_csv test case exhibits a convergence issue when running with -p 2 --n-threads 2, while it runs without issues using only -p 2.

Besides, recover failures are temporarily ignored for now.

Please let me know if you have any comments!

cticenhour · 2026-03-19T17:55:52Z

@lin-yang-ly I have been dealing with some similar brittleness for val-2i in #325. I might look at your PR for inspiration, though I think we're doing some similar things already in my most recent commits for that case. Might ask you some questions if I keep running into problems.

lin-yang-ly · 2026-03-19T18:04:29Z

@lin-yang-ly I have been dealing with some similar brittleness for val-2i in #325. I might look at your PR for inspiration, though I think we're doing some similar things already in my most recent commits for that case. Might ask you some questions if I keep running into problems.

Yes, just let me know if you have any questions!

simopier

Some minor comments and changes, and this additional request:

In the val-2c documentation, the 67.07% RMSPE value in the text for the calibrate results should be updated to 67.09% (not consequential, but still, let's keep it consistent).

loganharbour · 2026-03-19T21:37:50Z

@lin-yang-ly in case you didn't see it; https://civet.inl.gov/job/3673706/:

Failed Tests:
--------------------------------------------------------------------------------------------------------------
[3.075s] [ 204MB]     DIFF test:yttrium_hydrogen_system.YHx_PCT_T1273_P3e3_csv FAILED (CSVDIFF) [recover]

loganharbour

I'm still getting a diff with the petsc update (idaholab/moose#32404) + these changes:

test:val-2c.val-2c_delay_csv: Running csvdiff: /data/harblh/worktrees/moose/fix-cuda-typo/python/mooseutils/csvdiff.py /data/harblh/projects/tmap8/test/tests/val-2c/gold/val-2c_delay_csv.csv /data/harblh/projects/tmap8/test/tests/val-2c/val-2c_delay_csv.csv --relative-tolerance 3e-05 --abs-zero 1e-08
test:val-2c.val-2c_delay_csv: ERROR: In file val-2c_delay_csv.csv: The values in column "t2_paint_inventory" don't match.
test:val-2c.val-2c_delay_csv: 	relative diff:   9.033e-05 ~ 9.033e-05 = 6.019e-05 (6.019e-05)

Can we change that tolerance?

lin-yang-ly · 2026-03-19T22:44:50Z

I'm still getting a diff with the petsc update (idaholab/moose#32404) + these changes:

test:val-2c.val-2c_delay_csv: Running csvdiff: /data/harblh/worktrees/moose/fix-cuda-typo/python/mooseutils/csvdiff.py /data/harblh/projects/tmap8/test/tests/val-2c/gold/val-2c_delay_csv.csv /data/harblh/projects/tmap8/test/tests/val-2c/val-2c_delay_csv.csv --relative-tolerance 3e-05 --abs-zero 1e-08
test:val-2c.val-2c_delay_csv: ERROR: In file val-2c_delay_csv.csv: The values in column "t2_paint_inventory" don't match.
test:val-2c.val-2c_delay_csv: 	relative diff:   9.033e-05 ~ 9.033e-05 = 6.019e-05 (6.019e-05)

Can we change that tolerance?

Yes, we can change abs_zero align with case "val-2c_delay_calibrated_csv". Before that, I am curious that if the issue is from a local machine or on HPC? The reason I ask is I cannot reproduce the same issue on my local with current setting.

moosebuild · 2026-03-19T22:54:48Z

Job Build test summary, step Build test summary on ca367b7 wanted to post the following:

Test summary

Compared against c837769 in job civet.inl.gov/job/3665797.

No change

loganharbour · 2026-03-20T00:23:23Z

Before that, I am curious that if the issue is from a local machine or on HPC? The reason I ask is I cannot reproduce the same issue on my local with current setting.

This comes from the new version of PETSc (not available yet). So you won't be able to reproduce it. Once the new moose update goes through (it won't update in TMAP8 cause TMAP8 fails with it), I will update the moose submodule here and we can run it with the new moose environment.

simopier

This will be good to go if it passes the tests once @loganharbour adds the latest version of moose here.
Thank you!

(Ref. idaholab#380)

Co-authored-by: Pierre-Clement Simon <pierreclement.simon@gmail.com>

…issue

refs #0

loganharbour · 2026-03-20T22:39:59Z

@lin-yang-ly this has the new MOOSE submodule.

Before working with this branch (because I had to force push to bring things up to date), you should fetch from your own remote and reset to the new branch. You'll also need to update your conda environment to work with this. Let me know if you need any help.

lindsayad · 2026-03-23T23:30:01Z

we should just skip val-2f here as it will be fixed in #385

lindsayad · 2026-03-24T17:39:11Z

-    abs_zero = 1e-8
+    abs_zero = 2e-4


I think this is too big of a bump. Need some other way to make the test perform consistently

lindsayad · 2026-03-24T17:39:31Z

-    rel_err = 2e-5
-    max_parallel = 1 # see #200
-    recover = false # see #196
+    abs_zero = 2e-4


same comment

lindsayad · 2026-03-24T17:40:25Z

I would revert the large changes here and just skip for now while waiting on #385

lindsayad · 2026-03-24T17:40:56Z

+    cli_args = 'Executioner/nl_rel_tol=6e-6 InterfaceKernels/interface_reaction_YHx_PCT/silence_warnings=true' # should be removed once PCT curve implementation for low pressure is complete
    csvdiff = YHx_PCT_out.csv
-    max_parallel = 1 # see #200
+    recover = false # see #196


If we have to add this, then we're still not fixed

lindsayad · 2026-03-24T17:41:10Z

+                Executioner/TimeStepper/growth_factor=1.2
                InterfaceKernels/interface_reaction_YHx_PCT/silence_warnings=true' # should be removed once PCT curve implementation for low pressure is complete
    csvdiff = YHx_PCT_T1173_P5e4_out.csv
+    recover = false # see #196


shouldn't need

lindsayad · 2026-03-24T17:41:21Z

                InterfaceKernels/interface_reaction_YHx_PCT/silence_warnings=true' # should be removed once PCT curve implementation for low pressure is complete
    csvdiff = YHx_PCT_T1273_P3e3_out.csv
-    max_parallel = 1 # see #200
+    recover = false


shouldn't need

lin-yang-ly · 2026-03-24T17:46:16Z

we should just skip val-2f here as it will be fixed in #385

Sure, I will wait for the approval of #385. After that, I’ll move the current PR forward with updates of val-2f to align the performance between multi-cores and multi-OS.

lindsayad · 2026-04-10T16:06:49Z

Trying to understand why this was merged with over 2 week old testing? 😅

lindsayad · 2026-04-10T16:07:11Z

And with 2/11 tests passing

lin-yang-ly · 2026-04-10T16:25:55Z

Trying to understand why this was merged with over 2 week old testing? 😅

The work of this PR has been done in #385. Thus, we don't need to continue this PR again. I think the reason it shown this PR is merged is because the work of #385 is based on this branch.

lindsayad · 2026-04-10T16:32:11Z

So no one clicked the merge button on this PR? Well done tracking GitHub!

lin-yang-ly · 2026-04-10T17:47:40Z

So no one clicked the merge button on this PR? Well done tracking GitHub!

As I know, no one clicked it, and TMAP8 only have the merge history of #385.

simopier · 2026-04-10T22:14:08Z

That's right, I merged the other one and then came to this one to close it, only to find it "merged" already - by me.

I was even more surprised because the discussions are also not resolved, the tests were not passing... but since everything was addressed and fixed in #385, it's all good.

So Lin is right, Github made the link and closed this one too. I didn't even know it could do that.

lindsayad · 2026-04-11T23:32:06Z

well, all's well. Good work all

simopier self-assigned this Mar 11, 2026

simopier mentioned this pull request Mar 12, 2026

Val-2c/val-2f should perform consistently with 1 or 2 processors. #200

Closed

lin-yang-ly force-pushed the fix_failing_tests branch from 63270d1 to 2703079 Compare March 15, 2026 02:31

simopier requested changes Mar 19, 2026

View reviewed changes

Comment thread test/tests/val-2c/tests Outdated

Comment thread test/tests/val-2c/tests Outdated

Comment thread test/tests/yttrium_hydrogen_system/YHx_PCT.i Outdated

loganharbour requested changes Mar 19, 2026

View reviewed changes

simopier reviewed Mar 20, 2026

View reviewed changes

lin-yang-ly and others added 15 commits March 20, 2026 16:38

Fix light tests

9141968

(Ref. idaholab#380)

update val-2f HPC and local MacOS mismatch

54f0f4a

update petsc hyperparameters to solve the multi-cores mismatch

01adb18

update for multi-cores runs on HPC

745ddf4

avoid recover issue in this PR

562d5c3

update val-2c_immediate_injection_exodus for failure with 2 threads

3ad6b55

solve the multi-core issue for yttrium_hydrogen_system

303045c

keep multi-core issue of val-2f for future

8780a32

MacOS faulure of heavy tests will be solved in another PR

a1b7fee

move val-2f parameters to test file

9151669

leave the recover issue and val-2f multi-core issue for now

5cbba6d

Apply suggestions from code review

7d0c2f3

Co-authored-by: Pierre-Clement Simon <pierreclement.simon@gmail.com>

update test checks for solved recover and multi-core issue

150c67d

update documentation

3a765b6

solve one recover failure for yttrium_hydrogen_system

484f9a9

lin-yang-ly and others added 2 commits March 20, 2026 16:38

ignore recover issue due to current PR focusing on OS and multi-core …

9dae4e6

…issue

Update moose submodule

b0bf4d0

refs #0

loganharbour force-pushed the fix_failing_tests branch from 276589c to b0bf4d0 Compare March 20, 2026 22:39

simopier mentioned this pull request Mar 23, 2026

Extended PCT capabilities to low pressure for YHx #292

Open

solve failure for val-2c_delay_csv with moose update

0f0c001

lindsayad requested changes Mar 24, 2026

View reviewed changes

simopier mentioned this pull request Mar 30, 2026

Increase val-2f/val-2c reproducibility #385

Merged

simopier merged commit 0f0c001 into idaholab:devel Apr 10, 2026
2 of 11 checks passed

lin-yang-ly deleted the fix_failing_tests branch April 10, 2026 14:27

Uh oh!

Conversation

lin-yang-ly commented Mar 11, 2026

Uh oh!

moosebuild commented Mar 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

moosebuild commented Mar 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Coverage

Uh oh!

lin-yang-ly commented Mar 19, 2026

Uh oh!

cticenhour commented Mar 19, 2026

Uh oh!

lin-yang-ly commented Mar 19, 2026

Uh oh!

simopier left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

loganharbour commented Mar 19, 2026

Uh oh!

loganharbour left a comment

Choose a reason for hiding this comment

Uh oh!

lin-yang-ly commented Mar 19, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

moosebuild commented Mar 19, 2026

Test summary

Uh oh!

loganharbour commented Mar 20, 2026

Uh oh!

simopier left a comment

Choose a reason for hiding this comment

Uh oh!

loganharbour commented Mar 20, 2026

Uh oh!

lindsayad commented Mar 23, 2026

Uh oh!

lindsayad Mar 24, 2026

Choose a reason for hiding this comment

Uh oh!

lindsayad Mar 24, 2026

Choose a reason for hiding this comment

Uh oh!

lindsayad Mar 24, 2026

Choose a reason for hiding this comment

Uh oh!

lindsayad Mar 24, 2026

Choose a reason for hiding this comment

Uh oh!

lindsayad Mar 24, 2026

Choose a reason for hiding this comment

Uh oh!

lindsayad Mar 24, 2026

Choose a reason for hiding this comment

Uh oh!

lin-yang-ly commented Mar 24, 2026

Uh oh!

Uh oh!

lindsayad commented Apr 10, 2026

Uh oh!

lindsayad commented Apr 10, 2026

Uh oh!

lin-yang-ly commented Apr 10, 2026

Uh oh!

lindsayad commented Apr 10, 2026

Uh oh!

lin-yang-ly commented Apr 10, 2026

Uh oh!

simopier commented Apr 10, 2026

Uh oh!

lindsayad commented Apr 11, 2026

Uh oh!

Reviewers

Assignees

moosebuild commented Mar 11, 2026 •

edited

Loading

moosebuild commented Mar 11, 2026 •

edited

Loading

lin-yang-ly commented Mar 19, 2026 •

edited

Loading