After minor upgrade 8.17.0 to 8.18.15: less CPU but more 10 times more reads
Our burst credits went down to 0, because on AWS our app had started doing a lot more reading-actions than before. We noticed less CPU usage, which is good, but more 10 times more reads, which is very bad since it causes our burst credits to drop below 95. Even down to 0, causing a critical issue. One possible reason that we like to exclude: At the start of the many reading actions, one thing we implemented was a minor upgrade 8.17.0 to 8.18.15. Has anyone experienced something similar after a minor upgrade? Is this a known error?
Tim v Steenbergen
Did you check your long running microflows? Because you can see this kind of behaviour where you are in an endless loop of some kind.
We experienced a change in read/write and database CPU resource patterns after conversion from 9.6 to v13 of Postgres. This may or may not apply in your situation. My guess is that the query optimizer was making decisions according to new rules or based on an incomplete set of statistics.