Confirm 5.0 is not a memory hog 🐷, make necessary adjustments

[djbrooke]

In #7089/#7155 we added the ability to run memory tests on the 5.0 release candidate. We should now use this power to see if the release candidate is performing as well or better than 4.20 in terms of memory usage.

[mheppler]

Duplicate? Run Memory Tests on Release Candidate #7089

[pdurbin]

Per tech hours, @scolapasta will use this issue to coordinate the "human swarm." 🐝 🐝 🐝

#7155 is a good place to follow up on discussion of memory tests.

[pdurbin]

As discussed in tech hours, here are the results of the memory tests. The raw data and more is linked from the "Comparison for memory tests 4.20 vs 5.0 RC" spreadsheet at https://docs.google.com/spreadsheets/d/1Pghs_UIGb1QEmPl01HJ3T9KRvEZy619dp3dJ-2S6gTQ/edit#gid=0

4.20

4.20-withG1GC

5.0rc-78527a291

5.0rc-78527a291-noG1GC

[scolapasta]

Thanks for the graphs @pdurbin. As we discussed, these results are what we want to see - similar graphs when using the same GC. So for that purpose, these tests "passed".

One more test we still plan on doing is a team "swarm". Once we have the RC ready, we'll set an hour and everyone will go and use the system, create objects, etc. We'll keep this ticket open until we do that.

[scolapasta]

Swarm run. While we found an unrelated issue (with the publish button), nothing memory related jumped out as consequential related to 4.20. Closing this issue.