[Date Prev][Date Next] [Chronological] [Thread] [Top]

Re: Not able to apply patch....Directory missing...

>> In fact, that patch is for 2.1; if you use 2.1 you'll get far better
>> performances even without the patch (which is not yet released and
>> thus might result in an unstable system).
>> P.M.
> Thanx a lot for the reply. I'm quite new to LDAP.
> I have couple of other doubts, like
> - since i dont have the back-bdb how am i sure if am using berkeley
> DB(which is what i want to use). - Ant regarding 2.1 How unstable can
> the system get???

look at include/.portable.h after configure; you'll see
many #define/#undef lines; look at all those with DB in,
the HAVE_BERKELEY_DB will tell you if you're using the
Berkeley db.  About the version, you'll make configure
find the right one by providing appropriate CPPFLAGS
for the includes and LDFLAGS for the library

> Actually i'm trying to increase the search performance. And i was
> looking thru the open-ldap develop mailing list archives. I dint find
> much. Where (in the code) should i start. I was looking through the
> cache.c( in libraries/libldap/ directory). Theres a treshhold
> factor(TRESHOLD_FACTOR) and a SIZE_FACTOR, can i do something there?  I
> have indexed using slapindex, it does improve the performance, but not
> to a very large extent.

That has nothing to do with search performance, it's the
client-side cache, which is experimental and deprecated.
> I also have another doubt. Search tekes less time when i redirect the
> output of 'ldapsearch' to a file and not to a standard output(it sounds
> dumb) but what difference does it make. Does it mean that 'displaying'
> on the standard output is the cause for the delay. What i triend was
> display all the 1 lack entries i had it took almost 4.32 minutes. And
> when i redirected it to a file it took 2.46mins(!!!) i repeated it
> several times to make sure. The results are pretty much similar.  Coes
> it have any implication???? Please help me out.

The time your terminal wastes printig results has nothing
to do wit the performances of the server.  The figures you
show are reasonable, so please refer to client-server
interaction time; otherwise we better talk about the weather.


Pierangelo Masarati