In some of these tests, memory usage is low. We set minimum frag to 1.1, but when the memory is small, I think that we can fail to achieve this. The tests don't all have minimum frag levels which would ensure that 1.1 is meaningful. I saw locally a case where defrag just kept spinning and couldn't actually achieve 1.1. Another possibility is that it IS achieving 1.1, and defrag shuts off. But then a little new fragmentation is created during the INFO command which just marginally increases it again. Need to make sure that the success level is a little above the lower frag level to account for this.

I looked earlier this week to see if there were any open test issues, and didn't see any. Looks like we still have a few flaky ones.

I can dive deeper into any flaky ones after the holiday. If this is critical to eliminate the flaky tests now, I'd recommend just adjusting the success criteria a bit until we can do a deeper analysis.

Also, the one for "edge case" is a little weird. I have an open question to @zvi-code to get a second opinion. I'm not sure that the test is meaningful and maybe it should just be eliminated.

I can dive deeper into any flaky ones after the holiday. If this is critical to eliminate the flaky tests now, I'd recommend just adjusting the success criteria a bit until we can do a deeper analysis.

I wouldn't say it's critical to eliminate, I would just like the project to get to a healthier state after the holidays.

I might have a fix with #1492