Run failure when the tracer timestep exceeds the coupler timestep

[minghangli-uni]

In the 4ed0688, the ocean dynamic timestep and ocean tracer timestep are both set to 1800s, which is less than the coupler timestep of 3600s. Under this setup, the config runs successfully.

---

When the dynamic timestep is different from the tracer timestep which is increased to match the coupler timestep (for example, 3600s), the run proceeds as expected. However, if the tracer timestep exceeds the coupler timestep, the run fails with the following error in the access-om3.out,

3 total processes killed (some possibly by mpirun during cleanup)

This is followed by segmentation fault messages in access-om3.err,

[gadi-cpu-clx-2581:855569:0:855569] Caught signal 11 (Segmentation fault: address not mapped to object at address (nil))
[gadi-cpu-clx-2586:3332582:0:3332582] Caught signal 11 (Segmentation fault: address not mapped to object at address (nil))
[gadi-cpu-clx-2582:1539352:0:1539352] Caught signal 11 (Segmentation fault: address not mapped to object at address (nil))
[gadi-cpu-clx-2585:1181885:0:1181885] Caught signal 11 (Segmentation fault: address not mapped to object at address (nil))
[gadi-cpu-clx-2583:3307811:0:3307811] Caught signal 11 (Segmentation fault: address not mapped to object at address (nil))
[gadi-cpu-clx-2581:855544:0:855544] Caught signal 11 (Segmentation fault: address not mapped to object at address (nil))
...

This suggests that when the tracer timestep exceeds the coupler timestep, it triggers a segmentation fault, likely due to a mismatch in time synchronisation between components.

After a quick discussion with @ezhilsabareesh8 , a possible reason for the failure, as suggested by him, could be related to the fluxes being passed from WW3 to MOM6.

  ! wave to ocean coupling
  if ( associated(IOB%lamult)) then
    do j=js,je; do i=is,ie
      if (IOB%ice_fraction(i-i0,j-j0) <= 0.05 ) then
        fluxes%lamult(i,j) = IOB%lamult(i-i0,j-j0)
      else
        fluxes%lamult(i,j) = 1.0
      endif
    enddo ; enddo
    call pass_var(fluxes%lamult, G%domain, halo=1 )
  endif

I am unsure about this issue, so I'll leave it to you @ezhilsabareesh8 for your input and help.