Details
-
Improvement
-
Resolution: Duplicate
-
Major - P3
-
None
-
None
-
None
Description
I think we'd be better off setting up a new_handler that can print a stack trace and _exit() on out of memory since we can't really handle it well. Right now a bad_alloc error is often (amybe always) followed by segfault. Also we currently don't get a stack trace on bad_alloc because it unwinds the stack before we can catch it.
Attachments
Issue Links
- duplicates
-
SERVER-3612 Add a new_handler that aborts on bad_alloc
-
- Closed
-