MMS-spatial-fci test sporadically failing #3500

Activity

1. ZedThree

   added

   bug

   on Sep 21, 2026
2. ZedThree commented on Oct 8, 2026

   

   ZedThree

   on Oct 8, 2026

   MemberAuthor

   More actions

   Another failure, this time in hermitespline:

   Running command: ./fci_mms MYG=1 mesh:ddy:first=C2 mesh:paralleltransform:xzinterpolation:type=hermitespline mesh:file=mesh.64.fci.nc ... done in 1.61s
   ...
   div_par_K_grad_par errors: l-2 0.001415 l-inf 0.002827
   div_par_K_grad_par_mod_Original errors: l-2 inf l-inf inf
   div_par_K_grad_par_mod_ProductJK errors: l-2 0.001414 l-inf 0.002822
   div_par_K_grad_par_mod_Harmonic errors: l-2 0.001414 l-inf 0.002822
   ...
   

   So possibly just the ConductionMethod::Original that's failing? Although given that this is sporadic, that doesn't necessarily rule out the other methods
3. dschwoerer commented on Oct 9, 2026

   

   dschwoerer

   on Oct 9, 2026 via email

   Contributor

   More actions

   I managed to reproduce this locally. I think it was the first element of the not-first MPI thread. I am wondering whether there is a race somewhere, but I need further asserts in the code …

   …

   On 8 October 2026 5:33:59 pm GMT+01:00, Peter Hill ***@***.***> wrote: ZedThree left a comment (boutproject/BOUT-dev#3500) Another failure, this time in `hermitespline`: ``` Running command: ./fci_mms MYG=1 mesh:ddy:first=C2 mesh:paralleltransform:xzinterpolation:type=hermitespline mesh:file=mesh.64.fci.nc ... done in 1.61s ... div_par_K_grad_par errors: l-2 0.001415 l-inf 0.002827 div_par_K_grad_par_mod_Original errors: l-2 inf l-inf inf div_par_K_grad_par_mod_ProductJK errors: l-2 0.001414 l-inf 0.002822 div_par_K_grad_par_mod_Harmonic errors: l-2 0.001414 l-inf 0.002822 ... ``` So possibly just the `ConductionMethod::Original` that's failing? Although given that this is sporadic, that doesn't necessarily rule out the other methods -- Reply to this email directly or view it on GitHub: #3500 (comment) You are receiving this because you are subscribed to this thread. Message ID: ***@***.***>
4. ZedThree

   closed this as completedin #3510on Oct 14, 2026