Re: nslcd eats up all memory

[Ryan Steele <ryans [at] aweber.com>]

Hey Arthur,

Arthur de Jong wrote:
> On Sun, 2010-11-21 at 00:47 +0100, Peter Slickers wrote:
>> A bug causing similar symptomes has been uncovered and fixed recently
>> in PADL libnss-ldap. See Ubuntu's bugtracker for details:
>>
>> https://bugs.launchpad.net/ubuntu/+source/eglibc/+bug/292971
> 
> I think this is a different memory leak. So far I've tracked it down
> (using your feedback, thanks for that) to some memory that is not
> properly de-allocated when following referrals.
> 
> The relevant stack trace from valgrind is:
> 
> 16,736 (80 direct, 16,656 indirect) bytes in 1 blocks are definitely lost in 
> loss record 47 of 55
>    at 0x402328F: calloc (vg_replace_malloc.c:467)
>    by 0x4218CDF: ber_memcalloc_x (in /usr/lib/liblber-2.4.so.2.5.6)
>    by 0x408F0A0: ldap_send_server_request (in /usr/lib/libldap_r-2.4.so.2.5.6)
>    by 0x408FEB1: ldap_chase_v3referrals (in /usr/lib/libldap_r-2.4.so.2.5.6)
>    by 0x407B0A3: ldap_result (in /usr/lib/libldap_r-2.4.so.2.5.6)
>    by 0x804E6FE: myldap_get_entry (myldap.c:1059)
>    ...
> (the other memory leaks shouldn't grow during execution)
> 
> I can reproduce the leak in my test environment when using referrals. I
> have not yet narrowed it down to where the allocated memory should be
> freed though (perhaps in OpenLDAP code, perhaps in mine).
> 
> Perhaps a simple test case should be developed that demonstrates this
> leak that could be put forward to the OpenLDAP people.
> 
> Sorry this is taking so long.
> 
> 


As someone who has also observed this in production and is working on - thus 
far fruitlessly - debugging it, I just
wanted to say that the apology is appreciated but not necessary.  You have 
traditionally been very responsive to the
community, and respectfully cordial on top of it, and should be commended for 
your efforts instead of feeling obligated
to offer apologies.  Most of us using FOSS software know (or should, though 
that is unfortunately not always the case)
that we do so without any implied warranty or fitness for purpose, and it is 
quite admirable when FOSS project
authors/contributors put forth efforts equivalent to or exceeding those who 
would otherwise charge for them.

So, thanks for all your hard work.  If I manage to track the problem down in my 
endeavors I will be sure to pass the
information along on the list, provided someone else has not already done so.

Cheers,
Ryan

-- 
Ryan Steele
--
To unsubscribe send an email to
nss-pam-ldapd-users-unsubscribe@lists.arthurdejong.org or see
http://lists.arthurdejong.org/nss-pam-ldapd-users