Hasura graphql-engine crashes on accessing console

I deployed the Hasura docker container to an AWS t2.medium instance. It connects to an EC2 instance with a Postgres container. Initially it worked fine with one or two databases tracked. I added all databases and now the container crashes when attempting to access the console. Initially the console will display nothing for available tables in the schema. After a minute or two of trying to access the console the docker container crashes and brings down the AWS EC2 instance with it.

Docker logs outputs:

{"timestamp":"2019-10-16T16:42:38.742+0000","level":"info","type":"schema-sync-thread","detail":{"thread_type":"listener","info":{"received_event":{"instance_id":"cbf74cbf-774c-4e45-a8c5-31bc9b823f4a","occurred_at":"2019-10-16T16:43:20.632584Z"}}}}
{"timestamp":"2019-10-16T16:42:38.742+0000","level":"info","type":"schema-sync-thread","detail":{"thread_type":"processor","info":{"processed_event":{"instance_id":"cbf74cbf-774c-4e45-a8c5-31bc9b823f4a","occurred_at":"2019-10-16T16:43:20.632584Z"}}}}
graphql-engine: internal error: Unable to commit 1048576 bytes of memory
    (GHC version 8.6.5 for x86_64_unknown_linux)
    Please report this as a GHC bug:  http://www.haskell.org/ghc/reportabug

htop when it crashes:

[0x777]

@mzagaja Thanks for taking the time to dig through the problematic queries. We’ll improve these queries as soon as possible.

I think users might appreciate some kind of error that better suggests this in the logging when it fails for this reason, or a change that would lead to a more graceful failure than shutting down the instance

I’m not quite sure if we can catch this error but we should definitely document this behaviour.

On my end running a console refresh does not look like it is stressing our database server

The response of an introspection query is cached till the schema changes. This could be the reason.

[coco98]

Related to: #7585