Re: (ITS#8875) [Patch] Performance problems in back-mdb with large DITs and many aliases

[hyc@symas.com]

henrik.bohnenkamp@ionos.com wrote:
> On Thu, Jun 27, 2019 at 10:34:19PM -0700, Quanah Gibson-Mount wrote:
>> --On Sunday, April 07, 2019 10:15 AM +0000 henrik.bohnenkamp@ionos.com
>> wrote:
>>
>>> On Thu, Apr 04, 2019 at 05:32:16PM +0100, Howard Chu wrote:
>>
>> Had a customer who was hitting this issue try out these patches -- It
>> greatly decreases the search time (from unknown/infinite to 1 minute).
>> Unfortunately slapd then segv's.  Working on getting a test database to
>> reproduce the issue with for a good backtrace.
> 
> oh, that's too bad. I'd be happy to participate in the bug hunt. May I
> ask: which version did your customer use? I have modified the patch to
> work with the recent changes on master, but I did not dig in very deep
> to understand these changes. So I must admit that these modifications
> to the patch are completely untested (except the tests in the tests
> subdir).
> 
> Things might work better with the 2.4.46/47 patch, if that is an
> option for your customer.
> 
> Wrt getting a test-tree, maybe the python script I have in the github
> repo for the patch can provide
> one. (https://github.com/hbo/openldap-mdb-deref-problem/blob/master/scripts/write_tree.py)
> 
> I'm of course very interested to keep this patch alive, so if you have
> details on the segv, please let me know.

Fyi, on our problematic test database with 11M entries and 3.7M aliases, a search with -a always , starting from the
DB suffix, took 4 minutes without this patch, and 1235 minutes with this patch.

Needless to say, that's not looking good. Still checking other test cases.

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