search

jcpassy / mesa-h5

mesa hdf5 output se_support
BSD 3-Clause "New" or "Revised" License
2 stars 3 forks source link

restart problem #9

Open fherwig opened 5 years ago

fherwig commented 5 years ago

This problem was reported by Frank

porting SE support via a run_star_extra to the last public release, r10398, was not too difficult. just a few minor changes to the makefile_header.

i'm running a 12msun model.

porting SE support via a run_star_extra to the last public release, r10398, was not too difficult. just a few minor changes to the makefile_header.

however, a non-trivial issue has arisen.

with SE support in a run_star_extra, a job cannot be restarted. that is, any

% ./re photo

crashes immediately with the bizarre error: "PGSTAR Kipp failed to find star_mass in history data". the LOGS/history.data file has star_mass as one of its columns.

replace the SE run_star_extra with a non-SE run_star_extra and a restart happens normally, including the pgstar kipp.

this is a repeatable error on my 12msun model.

my sense is that some sort of memory stomping is going on that is triggered by a restart but not a run, and the memory stomp gives rise to a silly error message.

for now, i have turned off SE support as one goal is just to get a 12msun to core-collapse with a large-ish network.

jcpassy commented 5 years ago

I cannot reproduce this issue. I can restart a run with HDF5, with or without PGSTAR enabled. Have you been using the version on develop? I have issued a PR, please use this one. If the problem persists, send me files to reproduce the error.