maria-discuss team mailing list archive

[Michael Widenius <monty@xxxxxxxxxxxx>]

Hi!

>>>>> "Lance" == Lance Albertson <lance@xxxxxxxxxx> writes:

Lance> I've run into another weird problem where MariaDB is segfaulting
Lance> and dieing on Gentoo Hardened Linux. I can't figure out if its a weird
Lance> situation with my Gentoo environment on the machine or something with
Lance> MariaDB itself. I don't have this problem on the other machine running
Lance> MariaDB. It seems to happen under load about once or twice a day. I have
Lance> a nearly identical machine that doesn't have any of these problems.

Lance> I'm running mariadb-5.1.50 under Hardened Gentoo. The compiled flags are:

Lance> ---
Lance> Configuration summary for MariaDB Server version 5.1.50-MariaDB

<cut>

Lance> mysqld: malloc.c:5153: malloc_consolidate: Assertion
Lance> `nextchunk->fd_nextsize->bk_nextsize == nextchunk' failed.
Lance> 110329 13:11:25 [ERROR] mysqld got signal 6 ;

This looks like memory overrun, something that we have not seen in
MariaDB for a long time. (We are testing all builds with valgrind to
ensure we don't have any memory overruns).

<cut>

Is there a way we could install a debug binary on the machine in
question so that we could be able to debug this?

(We would need a binary compiled with debug + a core to be able to
find out what's going on).

The easyest way to get a debug binary is to get the source and
compile it with:

./BUILD/compile-pentium-debug-max

and then just copy it over your original mysqld binary.

If you can get us access to the system, we can help you with this.

You can find us at '#maria' channel on freenode if you want to discuss
this real time.

Regards,
Monty