Re: make test got stuck at Testing for slave slapd...

[Igor Brezac <igor@ipass.net>]

On Thu, 3 Apr 2003, Quanah Gibson-Mount wrote:

>
>
> --On Thursday, April 03, 2003 6:41 PM -0500 Igor Brezac <igor@ipass.net>
> wrote:
>
> >
> > On Thu, 3 Apr 2003, Quanah Gibson-Mount wrote:
> >
> >> Mark,
> >>
> >> Just an FYI... I've seen extremely poor performance results using Solaris
> >> 9.  Sun made a major change in the way in which Solaris handles threads
> >> between 8 and 9.  If you are looking to deploy a production OpenLDAP
> >> service, I suggest using Solaris 8 instead of 9.
> >>
> >
> > Hmm..  I do not find this to be correct.  I just ran a DirectoryMark
> > benchmark (100,000 entry dataset, 20 simultaneous clients, 100,000
> > different search samples) on Solaris 9 - fairly recent patches, (1 CPU
> > 440Mhz Ultrasparc IIi, 512MB RAM)  against Openldap 2.1.16 (tpool patch,
> > latest CVS SASL, Sleepycat 4.1.25).  I am able to get over 800 searches
> > per second and with more tweaking I can probably do better.
> >
> > This number is probably not indicative measure of a real world
> > application, but it does not indicate performance problems with Solaris 9.
> > I tested Solaris 8 awhile back and it performed roughly the same.
>
> Igor,
>
> I unfortunately have never seen the DirectoryMark's tool results have any
> resemblance to reality.  You gave me the same sort of answer when I was
> seeing massive problems with Solaris 8, with a rate  of 4-6 queries/second.
> After hiring Howard Chu to work on the system, in which he found the exact
> same results, an extensive set of patches were put in to OpenLDAP and
> related pieces of software (cyrus-sasl, heimdal, BDB, etc) before we got
> our current performance of 66/queries a second.  I have absolutely no faith
> in your results.
>

I respectfully disagree.

My setup is different than yours, one important difference is that I do
not use GSSAPI.  Binds to Openldap via SASL/GSSAPI are going to be
significantly slower than simple binds.  Another difference is how you use
LDAP, if all your connections to the LDAP server are 'persistent', this
particular DirectoryMark provides very valid test results.  In addition, I
believe some Openldap developers use DirectoryMark for Openldap
performance evaluation.

> What I do have faith in, is having the same software set up on two
> identical systems, with the exception of one running Solaris 9 with the
> latest patches, and the other running Solaris 8 with the latest patches,
> and running the same types of query sets against both systems.  The end
> results do not match, with Solaris 9 having a markedly worse performance
> result.  It has been about a month since I last did this test, and I'd be
> willing to do it again to see if any change has occurred, but it has been
> fairly consistent that Solaris 9's performance is worse.

How is this different from my claim?  I tested openldap systems on both
Solaris 8 and 9 and found virtualy no perfmance differnce between the two
operating systems.  I did compile packages on each OS separately.

I find your blanket statement of using one OS vs the other inapproriate
without providing detailed explanation of your situation.  A lot of
factors drive performance.  A good example is Linux where you'll find the
same version of the OS yielding different results.  In the end, all Mark
needed was help with a compile issue.

-- 
Igor