On 06/ 6/12 12:57 PM, Quanah Gibson-Mount wrote:

These may not be causing the issue you are seeing, but they should be fixed and then the setup retested. Of particular concern to me is item (a). I would make cn=monitor be olcDatabase {3}.

Done, thanks for pointing out the problems. I think I introduced them accidentally while backend-hopping during testing, but I'll check my prod setup as well.

I've made the changes and retested. The new node is still replicating, but after 50 cpu-minutes the process is at 10.2G and still going. I believe the leak is still present.