Comment 10 for bug 484779

Adding a stack traces for all threads when the system runs of out memory, as well as sequentials histogram of the heap over the run time of the performance test - (histogram sequencing was done manually while I was doing other things, so the intervals are not constant, but they do reveal certain trends)

You can see that the instance count and allocated memory for mondrian.rolap.RolapResult$CellInfo increases "linearly" over time, along with a few other bits that are probably related. My guess is that this is leaking somehow, however, we're not setup here to debug this correctly, not to mention we dont know this code. Can someone at T4Bi take a look? I'm getting a little tired of doing diagnostics on a black box with the cybernetic equivalent of a Ouija board as my tool set

thanks,