Description of problem: looks like results of IDXTLoad() never get garbage collected... test script included that shows it... just run that and it will eat about 25 megs every iteration... Version-Release number of selected component (if applicable): [root@sludge test]# rpm -q rpm rpm-4.2-0.67 How reproducible: pretty easy, see script... Steps to Reproduce: 1. run test script...
Created attachment 91183 [details] test script for showing memory leak
Heh, missed refcount on header while building tuple of tuples of headers. IDTXglob too. What's bogus (by design) is sucking every header out of a rpmdb into a >40Mb data structure. Getting an iterator on top of the glop, or adding cutoff time on retrieval, is gonna be needed. Changinging bug to RFE ...
refcount bug fixed, converting to RFE for iterator.
Moot problem afaik.