PETSc throws SIGABRT
This is a follow-up from #3194 (closed). I general, PETSc seems to have problems with the TRF process. (I think I was not able to get any example running using thermo_richard_flow. However, I could get some thermo_richards_mechanics examples running.)
For some cases (all project files in Tests/Data/ThermoRichardsFlow/SimplifiedMechanics
) it throws a SIGABRT:
[0] info: Solve processes.
[0] debug: Set the initial condition of variable temperature of process 0.
[0] debug: Set the initial condition of variable pressure of process 0.
[0] debug: SetInitialConditions ThermoRichardsFlowProcess.
[0] debug: Compute the secondary variables for ThermoRichardsFlowProcess.
[0] debug: Process output data.
[0] debug: process variable temperature
[0] debug: process variable pressure
[0] debug: secondary variable saturation
[0] debug: Computing new singular value decomposition
[0] debug: secondary variable velocity
[0] debug: output to TRuni_unsat_ts_0_t_0.000000.vtu
[0] debug: Writing output to 'TRuni_unsat_ts_0_t_0.000000.vtu'.
[0] info: [time] Output of timestep 0 took 0.00324557 s.
[0] info: === Time stepping at step #1 and time 0.1 with step size 0.1
[0] info: Calculate non-equilibrium initial residuum.
[0] debug: Assemble the Jacobian of ThermoRichardsFlow for the monolithic scheme.
[0] info: [time] Assembly took 0.0289182 s.
malloc(): unaligned tcache chunk detected
[modmon160:83187] *** Process received signal ***
[modmon160:83187] Signal: Aborted (6)
[modmon160:83187] Signal code: (-6)
malloc(): unaligned tcache chunk detected
[1] 83187 abort (core dumped) ~/github/ogs/build_mkl_petscevennewer_debug/bin/ogs TRuni_unsaturated.prj
Description
[Description of the bug or feature]
Steps to Reproduce the Problem
- It happens e.g., if
<petsc><parameters>-ksp_type cg -pc_type bjacobi -ksp_rtol 1e-16 -ksp_max_it 10000</parameters></petsc>
is added to the linear_solver section ofTRuni_unsaturated.prj