Hi Peter,
On Fri, Apr 15, 2005 at 05:58:17PM +0200, Petr Vandrovec wrote:
slapd/slapadd use sched_yield on places where it really should not be idling. Due to this upgrade from 2.1.30 to 2.2.23 takes more than 1 day on our 100000 objects tree
Erm. Wow!
we have, with about 40 indexed attributes (after one day
I gave up, attached debugger to slapadd and overwrite
sched_yield procedure in slapadd with 'ret'; import
finished ~1 hour after I did that).
Nice. :) Could you please check if the new slapadd -q option in 2.2.23-2 improves the situation for you?
I've tried it on my system at home, and it indeed improves situation a lot - so I think that you can close this bug, slapadd -q seems sufficiently cut time down when box is idle during import.
Only strange thing is relation between dbsync/dbnosync of ldbm backend and time to perform import. With dbnosync it took 2h04m18s to import data with 'slapadd -q', with 'dbsync 50 30 1' it took 50m35s, with 'dbsync 5 30 1' 39m30s, and with 'dbsync 1 30 1' it took 1h27m12s. User time consumed was always 2m34s-2m35s, and system time 2m2s-2m10s. So even in best case it spent 35 minutes of import time by waiting :-(
Thanks,
Petr Vandrovec
-- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
