Fix: pengine: Make sure calculated resource scores are consistent on different architectures
Previously, pengine test "bundle-nested-colocation" failed on x86
architecture:
@@ -334,8 +334,8 @@ native_color: rabbitmq-bundle-docker-0 allocation score on overcloud-galera-1: -INFINITY native_color: rabbitmq-bundle-docker-0 allocation score on overcloud-galera-2: -INFINITY native_color: rabbitmq-bundle-docker-0 allocation score on overcloud-rabbit-0: INFINITY -native_color: rabbitmq-bundle-docker-0 allocation score on overcloud-rabbit-1: -9999 -native_color: rabbitmq-bundle-docker-0 allocation score on overcloud-rabbit-2: -9999 +native_color: rabbitmq-bundle-docker-0 allocation score on overcloud-rabbit-1: -9998 +native_color: rabbitmq-bundle-docker-0 allocation score on overcloud-rabbit-2: -9998 native_color: rabbitmq-bundle-docker-0 allocation score on rabbitmq-bundle-0: -INFINITY native_color: rabbitmq-bundle-docker-0 allocation score on rabbitmq-bundle-1: -INFINITY native_color: rabbitmq-bundle-docker-0 allocation score on rabbitmq-bundle-2: -INFINITY @@ -347,7 +347,7 @@ native_color: rabbitmq-bundle-docker-1 allocation score on overcloud-galera-2: -INFINITY native_color: rabbitmq-bundle-docker-1 allocation score on overcloud-rabbit-0: -INFINITY native_color: rabbitmq-bundle-docker-1 allocation score on overcloud-rabbit-1: INFINITY -native_color: rabbitmq-bundle-docker-1 allocation score on overcloud-rabbit-2: -9999 +native_color: rabbitmq-bundle-docker-1 allocation score on overcloud-rabbit-2: -9998 native_color: rabbitmq-bundle-docker-1 allocation score on rabbitmq-bundle-0: -INFINITY native_color: rabbitmq-bundle-docker-1 allocation score on rabbitmq-bundle-1: -INFINITY native_color: rabbitmq-bundle-docker-1 allocation score on rabbitmq-bundle-2: -INFINITY
In this case, given:
float factor = 10000.0 / 1000000;
int score = -1000000;
int weight = factor * score;
, the "weight" would be -10000 on x86_64, but -9999 on x86.
This commit fixes it by rounding the number according to the description
in http://c-faq.com/fp/round.html:
"Note that because truncation is otherwise the default, it's usually a
good idea to use an explicit rounding step when converting
floating-point numbers to integers. Unless you're careful, it's quite
possible for a number which you thought was 8.0 to be represented
internally as 7.999999 and to be truncated to 7."