RE: Segfault during Auth Benchmark request

["SAGNIMORTE Thomas (CAMPUS)" <thomas.sagnimorte@oxylane-group.com>]

Hi,

I am using hdb backend.

Here is the gdb backtrace 
Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 1140881728 (LWP 31250)]
0x0000003c024082e9 in pthread_mutex_lock () from /lib64/libpthread.so.0

I set entry cache to 100000, idlcache to 10000 and dncache to 10000.

Regards,

Thomas

 

-----Original Message-----
From: Quanah Gibson-Mount [mailto:quanah@zimbra.com] 
Sent: Tuesday, April 21, 2009 6:58 PM
To: SAGNIMORTE Thomas (CAMPUS); openldap-technical@openldap.org
Subject: Re: Segfault during Auth Benchmark request

--On Tuesday, April 21, 2009 6:24 PM +0200 "SAGNIMORTE Thomas (CAMPUS)" 
<thomas.sagnimorte@oxylane-group.com> wrote:

>
> Hi,
>
> I am working on building and benchmarking OpenLDAP v2.4.16 for a large

> entries directory (actual test: 2 000 000 but final objective 10 000 
> 000 users).
>
> I use SLAMD to test authentication in order to evaluate what is the 
> maximum number of authentication I can do on my server. So I try to do

> 5 authentications per 5ms during 5min.
>
> My OpenLDAP 2.4.16 is install on the X86 two core 2.8Ghz server with 
> 4Go of memory.
>
> I set 3Go in cache of DB_Config and I have look a lot on how configure

> properly slapd.
>
> Is it normal because I have reach system limits or can I change some 
> parameter to optimize slapd performance?
>
> Do some people have some figures about slapd performance with large 
> entries directory on standard server?

Where is the gdb backtrace of your segfault?

Which slapd backend are you using? bdb? hdb? something else?  Did you
properly tune the entry cache, idlcache, and dncache?

--Quanah

--

Quanah Gibson-Mount
Principal Software Engineer
Zimbra, Inc
--------------------
Zimbra ::  the leader in open source messaging and collaboration