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

Re: Another observation about the beta release ...

Thanks for the hints. After attempting the same import listed below a couple different ways, we attempted it using an LDIF.txt file and the ldapmodify tool. It worked perfectly without a hitch, and the server didn't experience the excessive memory use (though it did top out at about 160 meg). What is truly baffling is we ran the ldapmodify command on the SAME machine as the slapd server was running on, unlike the perl program (which was run on a remote machine across the network).

I will hopefully have some time to look into this tomorrow. On another note, I was wondering how we're supposed to add indexes after creating the database. I would LOVE to use the utilities mentioned in the faq (and which I have used in the past), but these utilities apparently are no longer included with the development version. Are they planned for the future, or has the functionality been merged into something else?

Overall, I am very impressed so far. The only other draw back has been performance. A simple search against the 1.2.11 version returns (across the network) in less than a second (using an identical database with identical indexes). The newer version returned the same results in approximately 7 seconds under identical conditions. Once I tweaked the slapd daemon using the 'cachesize' and 'dbcachesize' options, the response time after the first query was reduced to 3 second.

The main differences (which I am aware of) between the two servers are the version and that the version 2 slapd has fewer ACL's (I deleted a few to see if it would improve performance, which it didn't appear to make any difference). If there is anything I can do to help troubleshoot this definitively, please let me know.

Tony

At 10:50 AM 7/14/00 -0700, Kurt@OpenLDAP.org wrote: 
>>After doing part of the transfer, we noticed the slapd server occupied
>roughly 41 meg of memory (following the import of a 100 or so entries).
>I then killed the slapd server, and restarted it.  It only occupies
>roughly 1 meg after restart (the same amount before we imported the
>information.  We then added roughly have the previous number of entries
>after restart, and once finished, the server now occupies 49 meg of
>memory.

Note that slapd uses caching (of entries and indexes) and may not
reach a stable memory usage for quite some time.


******************************************************************************
* Anthony Brock                                         abrock@georgefox.edu *
* Director of Network Services                         George Fox University *
******************************************************************************