osisaf drifters hang when restarting

[tdcwilliams]

accidentally put this in wrong repo
https://github.com/nansencenter/nextsim-env/issues/54

[tdcwilliams]

thanks for this comment @fsalmon001, it should help:

For me the issue should stem from PR 678. I did not have any OSISAF files so I was not able to test them.

However, I think this could stem from the initialising function in drifters.hpp. I had a similar issue. Some processes or just rank 0 I don't remember returned true and the other(s) false, and therefore the process hung at an MPI function. You can check this with print functions.

If so, the problem could be an issue with the current_time and M_time_init. Because of the restart, maybe the M_init_time is not correct I don't know. I faced this kind of issue for the SPACING option but I solved it, maybe not completely.

[tdcwilliams]

Hi @fsalmon001
I stopped the hanging with this commit

However the M_rank > 0 drifters are not initialised. Probably this could be done in the place where I change M_time_init with setInitTime, but I am not clear on how to do it with parallel drifters. Could you suggest a change for that line?

[fsalmon001]

Hi @tdcwilliams, what you have done sounds good. I do not think we need more because the drifter trajectories are still stored on rank 0 only. Does the simulation fail later? Is there still an issue with drifters?