While running a file which was too big for the GPU we were using, the notebook crashed. After reducing file size and restarting the notebook, running the job_init_pass() crashed the kernel, which then required running the script from the start. As there was a successful initial pass before, the file was not to blame.
When looking at the share memory at /dev/shm/ the first file was still loaded in memory. Clearing the memory with rm 'shared_memory_id' solved the kernel crashing issue.
closing this since it's platform-specific and is fixed with a reboot or clearing the memory. though will try to reduce shared memory usage in the future
While running a file which was too big for the GPU we were using, the notebook crashed. After reducing file size and restarting the notebook, running the job_init_pass() crashed the kernel, which then required running the script from the start. As there was a successful initial pass before, the file was not to blame.
When looking at the share memory at /dev/shm/ the first file was still loaded in memory. Clearing the memory with rm 'shared_memory_id' solved the kernel crashing issue.