The current problem in troubleshooting of OOM incidents often boils down to figuring out which operations were responsible for excessive memory allocations.

Very often even with knowing the exact operations that were executed at the time of the incident, the memory requirements of those can only be speculated about but never known with a greater degree of confidence.

This presents a pain point for the support team.

This ticket is raised to explore if it is possible to log memory allocations if those happen as a result of an execution of a particular operation.