Memory leak in LegacyLoopTask, line 54

[jwalcorn]

If you run enough iterations via loopctl, you'll eventually get an OutOfMemoryError. For example:

This appears to be because the loopResults list grows without bound, getting a bit larger each iteration. Since we're holding on to this outside the loop, it can't get garbage collected. So line 54 of LegacyLoopTask.java means we eventually run out of memory, if asked to do enough iterations

[jwalcorn]

[rtclauss]

I'll take a look at adding JDBC/SQLite support to store the results

[jwalcorn]

OK, cool. Though note this client is just a big Java main(), not running in an app server like Liberty or Tomcat or Quarkus. So that limits things a bit (like connection pooling, etc), even though it is possible to use JDBC/JPA outside of an app server, of course.

[jwalcorn]

I was thinking a bit more about this. I think we should probably just keep track of aggregates, rather than a full recording of the results of each and every iteration (whether in memory or in a database). For example, we could keep a HashMap (or whatever) where the key is the error message (or http return code), and the entry is a counter we could increment each time we encounter a certain error. Then at the end we could show a success percentage, and the top 10 (or whatever) errors that occurred during the run.