STUCK at ---[precice] Synchronize participants and close communication channels

Facing problem while running the turek-hron-fsi3 tutorial. Dealii is stuck at —[precice] Synchronize participants and close communication channels. Please have a look at the screenshot. Please guide me on this. Thank you!

Hi @Piyush_Pulukool,

I understand that on the deal.II side, preCICE initializes and then immediately finalizes. This could mean that something went wrong on deal.II in the meantime and it is trying to exit.

Have you changed anything regarding the tutorial?

Maybe @DavidSCN can think of potential issues here.

I guess the overall issue is described here C++ throw leads to a deadlock in preCICE · Issue #964 · precice/precice · GitHub. Problem is that somewhere in the deal.II code, an assertion is thrown, which calls the preCICE destructor. That’s a problem of preCICE’s error handling.
However, the important question is why if fails. @Piyush_Pulukool did you modify the tutorial in some way?

@Makis o, @DavidSCN I have not modified any part of the source code. I am following the tutorial.

Hi everyone! Just need a follow up on this one. Am I the only one facing this problem? Should I reinstall any packages? All Dealii cases seem to have the same problem in my system. OpenFOAM - CalculiX cases run fine with no issues.
Thank you for your time.

Distinguished scholar. Have you solved this problem?

Hello, I solved this problem just now. If you are using a linear solver, please be sure to change it to CG instead of Direct.

1 Like

In case that’s the problem: You need to compile deal.II with UMFPACK, which provides the direct solver. If you compile deal.II without UMFPACK and try to run any case using the direct solver, deal.II will throw an error, which can indeed lead to a deadlock (see linked issue above). At the moment, I don’t see any obvious way to prevent this behavior (unless we want to restrict the adapter usage only to deal.II versions having UMFPACK). Note that the direct solver is most likely multiple times faster (depending on your problem size).

1 Like