Add setting to avoid memory leak, and add comment about number of workers
Steps taken to alleviate memory-related performance issues and errors occurring since the Ginkgo upgrade.
Note that 80000 is an arbitrary number that should restart workers every few days. With 3 LMS this gives 240k requests, +10% static 260k, this means that only after 260k requests will a worker restart, this can be 6 days if 40k requests/day, 12 days if 20k requests/day, etc.
An alternative solution would be to do this change in the configuration
branch instead of Ocim.
Testing
- deploy an appserver in an instance that has already appservers
- check that the new setting has been correctly applied to the new appserver (but not changed in the old ones)