The 3.2 release notes upgrade procedure recommends disabling core dumps in production:

For production environments, you may prefer to turn off core dumps for the operating system, if not already.

Core dumps provide information which can be invaluable in situations where a mongod/mongos has experienced a hard crash. In cases where the crash happens very infrequently, with unknown cause/triggers, and so cannot be reproduced, having a core dump from an unexpected event may be the difference between fixing the bug (preventing the possibility of recurrences) and not. Keeping core dumps on production systems is a useful practice in much the same way that logs of prod systems are useful.

DOCS-6018 alludes that "[core dumps] may cause problems in production environments", but does not explain what those perceived problems are seen to be. ramon.fernandez, can you please elaborate on this point?