I am currently using Postgres hosted on Heroku and Hasura for GraphQL. When having slightly higher amount of active users (~100) we are experiencing enormous connection issues. Looking at the heroku logs it says "sql_error_code = 53200 DETAIL: Failed on request of size 224 in memory context "MessageContext"."
However, the metrics show that we should have enough RAM (max used 33%) and dynos.
Is this a problem of connections between postgres and hasura? Can it be caused by a certain expensive Query? What other configurations should I look at?