the issue here is that the cache loading was sped up by tweaking how much data was loaded as well as the number of roundtrips required to pull that data into the cache. as a result, AlertCondition had two fields marked as lazy, which were previously marked as eager. i missed this during my initial testing, because i was focusing on the performance bottleneck at the time: measurement-based alerts. and our smoke testing missed this too.
branches/RHQ_1_0_1_GA_CP : rev1274 - only eagerly load the condition's corresponding MeasurementDefinition, if this is a measurement-based condition;
trunk (1.1.0-SNAPSHOT) : rev1276 - only eagerly load the condition's corresponding MeasurementDefinition, if this is a measurement-based condition;
Tested an operation based and availability based alert, both worked and could view in the history. rev1304, linux, postgres.
This bug was previously known as http://jira.rhq-project.org/browse/RHQ-751