The current method of polling the DockerHub URL uses a proxy provided for free by an OS project, but the connection is rate limited to 200 connections per hour. This means we can exceed out limits if lots of people are using the site, in which the data will not load for an hour.
One solution is moving the polling to a script that Travis runs and then saving that data to a JSON file, and using that to create the views. This could be ran once a day as a CRON job so the data is never more than 24 hours old.
We could also take an OS proxy server and deploy it to IBM Cloud, where we won't hit any rate limits as we will own the instance.
The current method of polling the DockerHub URL uses a proxy provided for free by an OS project, but the connection is rate limited to 200 connections per hour. This means we can exceed out limits if lots of people are using the site, in which the data will not load for an hour.
One solution is moving the polling to a script that Travis runs and then saving that data to a JSON file, and using that to create the views. This could be ran once a day as a CRON job so the data is never more than 24 hours old.
We could also take an OS proxy server and deploy it to IBM Cloud, where we won't hit any rate limits as we will own the instance.