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

Re: hdb vs. search base



Cache sizes should not make the difference between correct and incorrect results. Sounds like you've found a bug that should be filed in the ITS. It would be good if you can provide a minimal database that demonstrates the problem as well.

Donn Cave wrote:
Testing an organization white pages directory with 2.3.21
and hdb backend, I was disappointed to find that some
search bases don't work -- 0 entries found -- where more
general or more specific bases do work.  Is this an expected
result, maybe something that depends on tuning parameters
like idlcachesize?

The structure is, sort of diagrammatically,
   suffix
       people
           students
                dept1
                    student1
                    ...
                dept2
                ...
           staff
                dept8
                    staff1
                    ...
                dept9
                ...

So far, it looks like the following search bases work:
   suffix
   students,people,suffix
   deptx,students,people,suffix
   deptx,staff,people,suffix

and the following do not work:
   people,suffix
   staff,people,suffix

"staff" contains about 6800 departments.  It's garbage that
people type into web forms and whatnot, every possible spelling
of every departmental unit, and punctuation tends to be somewhat
haphazard too.  It builds without errors, though.

I can use bdb for this, works fine, but it would be interesting
to find out where hdb is going wrong, if anyone can think of
further useful experiments.  So far I have tried idlcachesizes
from none to 10000.

Thanks,

    Donn Cave, donn@u.washington.edu






--
 -- Howard Chu
 Chief Architect, Symas Corp.  http://www.symas.com
 Director, Highland Sun        http://highlandsun.com/hyc
 OpenLDAP Core Team            http://www.openldap.org/project/