Review Webservice pod memory recommendations
Inspired by this customer ticket (internal link) (and also possibly this customer ticket (internal link) I wanted to discuss raising the default webservice memory limit uses.
Currently it is set to 5.25G. Since GET by default uses 4 CPUs our documentation recommends 7G as a maximum for the limit.
When under higher load this ends up causing hard to diagnose 502 errors as individual Puma workers get killed but the pod itself does not get killed or restart.
Edited by Grant Young