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

Re: syncrepl w/ 2.3 and bdb 4.3 working great

--On Monday, September 12, 2005 9:58 AM -0400 Dusty Doris <openldap@mail.doris.cc> wrote:

How is your testing doing with OL 2.3.7 and BerkeleyDB 4.3.28?

I was wondering if BerkeleyDB 4.3.28 was suitable for production use
with OL 2.3.7.

So far, things are looking good with 2.3.7 and bdb 4.3.28. For my recent test, I did a dump on my production directory and imported it into my lab machine. That gave me about 300,000 entries. Then I ran a script that did the following:

-add 10,000 users
-delete 5,000 users of those new users
-add 5,000 of those users back and 5,000 additional users
-modify 10,000 of the 15,000 new users

That was run with a single ldapmodify command that took all the changes
from a file.  I then ran searches against the master and two slaves while
the adds were going.  The searches will still fast during all the adds.

At the end of the script, I did an ldapsearch and compared the contextCSN
numbers w/ master and the two slaves.  They both checked out.

Then I did an ldapsearch objectclass=* under the base of my users on each
server, output to a file (> ldap${host}.ldif) and then ran a diff on
those files.  All the files were the same, no diff.

So, it looks good to me so far.  I have my development provisioning
system now pointing to the development ldap server and all customer
adds/deletes/modifications are flowing through correctly and replicating

I'm not seeing any issues, but then again, I'm not working with a whole
lot of data here either.

The problems I had with BDB 4.3 (21 & 27) were related to data loading via slapadd (and the use of IN-MEMORY logs). The "-q" function in 2.3 removes the necessity of the IN-MEMORY logs. The other issues seen in 21 & 27 were reported by other people, and may well have been resolved in 4.3.28.


Quanah Gibson-Mount
Principal Software Developer
ITSS/Shared Services
Stanford University
GnuPG Public Key: http://www.stanford.edu/~quanah/pgp.html