Open micahjohnson150 opened 4 years ago
I'm working through this as well. However, we have to specify what happens when a restart is needed. I think that throwing out depth is a last resort and the thresholds should be changed first to see if it can be made through.
After chatting with @etrujil who is running awsm right now, a crash restart should be an automated thing if the restart_crash = True. As of now (I think) the restart also requires a wyh_restart_output to restart at, but its description says its the last output prior to the crash which seems to me to be determinable. By having this automated it will certainly make us more reproducible and by having the control crash_restart we aren't losing control.
Coreconfig description https://github.com/USDA-ARS-NWRC/awsm/blob/dd323c09109a4bcc470d84e35e7d96e239fea074/awsm/framework/CoreConfig.ini#L238
Location of the wyh_restart in framework.py https://github.com/USDA-ARS-NWRC/awsm/blob/dd323c09109a4bcc470d84e35e7d96e239fea074/awsm/framework/framework.py#L774