search

prior-art-archive / query-parser-docker

Dockerized version of the parser (starting w v1)
1 stars 0 forks source link

Setup hosting for this image #10

Open reefdog opened 5 years ago

reefdog commented 5 years ago

Let's get this image hosted (on Heroku? Amazon ECS?) so we can interact with it as our query parser endpoint rather than the Lambda invocation.

slifty commented 5 years ago

It looks like Amazon ECS is the mechanism for hosting docker containers on AWS; I'm going to look into what it takes to do correctly and document here.

We may want to consider setting up a VPC (Virtual Private Cloud) so that nobody can directly access / query the docker container, and only the prior art archive's backend has access. However, this would require eventual migration of the entire infrastructure onto AWS so that's not really an option for now.

We can also host this dockerized parser on Heroku which may be the right solution given the way Heroku is the core platform for most of the infrastructure.

slifty commented 5 years ago

The v1 deployment is currently on Heroku; I believe at docker-usptofe

There appears to be a memory leak somewhere in the application

image

I want to confirm my understanding of the deployment before spinning this out into its own issue.