Hi Yan,
I tested the correction.
The problem that I reported was improved by the correction that you performed.
When I tested it, I found a different problem.
When I set "placement-strategy=default", it is a problem that a function to
disperse does not commit placement by the number of the start resource.
When this problem applies the patch which I affixed it to, it is settled.
Is the correction with this patch right?
Best Regards,
Yuusuke IIDA
(2011/04/27 13:32), Yan Gao wrote:
Hi Yuusuke,
On 04/19/11 19:55, Yan Gao wrote:
Actually I've been optimizing the placement-strategy lately. It will
sort the resource processing order according to the priorities and
scores of resources. That should result in ideal placement. Stay tuned.
The improvement of the placement strategy has been committed into the
devel branch. Please give it a test. Thanks!
Regards,
Yan
--
----------------------------------------
METRO SYSTEMS CO., LTD
Yuusuke Iida
Mail: iiday...@intellilink.co.jp
----------------------------------------
diff -r f59d7460bdde pengine/allocate.c
--- a/pengine/allocate.c Thu Apr 21 14:59:43 2011 +0200
+++ b/pengine/allocate.c Thu May 12 19:30:01 2011 +0900
@@ -989,17 +989,14 @@
stage5(pe_working_set_t *data_set)
{
GListPtr gIter = NULL;
+ GListPtr nodes = g_list_copy(data_set->nodes);
- if (safe_str_neq(data_set->placement_strategy, "default")) {
- GListPtr nodes = g_list_copy(data_set->nodes);
+ nodes = g_list_sort_with_data(nodes, sort_node_weight, NULL);
- nodes = g_list_sort_with_data(nodes, sort_node_weight, NULL);
+ data_set->resources = g_list_sort_with_data(
+ data_set->resources, sort_rsc_process_order, nodes);
- data_set->resources = g_list_sort_with_data(
- data_set->resources, sort_rsc_process_order, nodes);
-
- g_list_free(nodes);
- }
+ g_list_free(nodes);
gIter = data_set->nodes;
for(; gIter != NULL; gIter = gIter->next) {
_______________________________________________
Pacemaker mailing list: Pacemaker@oss.clusterlabs.org
http://oss.clusterlabs.org/mailman/listinfo/pacemaker
Project Home: http://www.clusterlabs.org
Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
Bugs: http://developerbugs.linux-foundation.org/enter_bug.cgi?product=Pacemaker
