Things have been nothing less than crazy over the last two months. We have gone from 300.000 uploads to almost 1.000.000. During peaks, we are now fully utilizing 20 machines (160 cores), and we are still continuing to grow quickly.

Unfortunately, this also meant that a lot of our time went into scaling the service, rather than building shiny new things. As of today, we have finally caught up with some of the most pressing issues, allowing us again to allocate time towards other improvements.

To start with, we have made a few small improvements to our Assembly detail page. Rather than dumping a huge JSON object at the bottom of every page, we are now rendering all aspects of an Assembly in a more useful way.

This includes listing all uploads and results in a table like this:

With detailed information showing up in a modal overlay:

While we were at it, we also improved the information we provide in cases where there was an error while executing the Assembly:

To make this change even sweeter, we also migrated to a faster database server once again, after an EC2 failure caused us to move to a weaker fail-over machine for a while. As opposed to our last announced downtime, which lasted 4 minutes, we were now able to perform the switch without any downtime whatsoever! 😄.

The Assembly list page now also shows the used redirect_url and notify_url.

Let us know if you like the changes!