Need help with jemalloc cores
jasone at canonware.com
Wed Sep 14 16:34:42 PDT 2016
On Sep 13, 2016, at 10:57 PM, Asmita Jagtap <ajopensrc at gmail.com> wrote:
> Forgot to mention the version, it is jemalloc 4.2.1
> On Wed, Sep 14, 2016 at 11:11 AM, Asmita Jagtap <ajopensrc at gmail.com> wrote:
> We are using jemalloc library (compiled with --enable-fill and --enable-debug options) in a multi-threaded daemon and seeing multiple cores during malloc/realloc with jemalloc library. The core files are approximately 61GB and 79GB in size, so this is typically seen under heavy memory usage situation.
> Are these issues in jemalloc code or the application code that is using jemalloc?
It's hard to say for sure whether this is an application bug or jemalloc bug. Is it possible that the application is double-freeing a huge allocation? If you're unable to narrow down the problem, I recommend running with a debug build of jemalloc to see if any assertions fail. Hopefully that will help us determine where things first go wrong.
P.S. I'm just about to retire this mailing list, so if you need to follow up on this thread, please create a new GitHub issue (https://github.com/jemalloc/jemalloc/issues), and refer to this thread's archive (e.g. http://jemalloc.net/mailman/jemalloc-discuss/2016-September/001326.html).
More information about the jemalloc-discuss