Web Systems Release Notes: 2017.45

Could not scroll through slideshow images or expand the slideshow to full screen.

A search for Heart Transplant resulted in the article titled "Heart Transplant" displaying second in the search results, not first. We have added a relevance-based sorting rule that addresses this. These were previously sorted alphabetically.

Updated the hero image on the Uveitis organization page (https://www.yalemedicine.org/departments/uveitis/) to provide a better cropped image.

The doctor filter on the Condition articles page and the Article Reporting Tool now show Concept Articles for a given Doctor using the same logic as that used on the public-facing YM site.

The doctor filter on the Stories page now shows Stories for a given Doctor using the same logic as that used on the public-facing YM site.

If a Story is given a Hero Type of Video, its Type will be automatically switched to Video if necessary.

Thumbnail images were not appearing for some audio players.

Calendar editors were receiving daily suggested event notifications for events which had been deleted.

Internal system change that was necessary for the "Deep Brain Stimulation" accessibility fix in TECH-2705 (In Review) .

This accessibility fix labels a link to a Condition article differently from a link to an Organization that happens to have the same name – e.g. "Deep Brain Stimulation", which is both a procedure and an organization.

Adjusted client-side logic to show alternative text for promo images.

Added required data in the Search index to show alternative text for promo images.

DNS entry pointed to an old server and had to be changed by Central ITS.

Fixed: Central ITS changed the IP addresses used.

Currently the Print Publications website points to the preview broker database for SDL because the live broker breaks when used. This covers the work necessary to create a separate broker database dedicated to Print Publications as a medium term fix until a long term project to overhaul Print Publications is completed.

Our search indexes are currently served from a single location in the USA, this task is to create globally redundant search indexes to improve application reliability.

When users publish to live in SDL we also force a publish to preview so that preview content matches what is in live even if the user does not check the "preview" box when publishing. Now that the Print Publications runs on its own broker with separate Live publishing, we need to enable this feature to prevent preview and live getting out of sync if someone was to only publish to live.

Database was made globally redundant with a failover instance running in Northern Europe.

This task changes the URL used for presenting the thumbnail for each video.

This task is to change the video player to work with Vimeo content that was converted to Vimeo from Technolutions.

This changes the SDL Content Delivery service used to get content for Print Publications to the new dedicated broker.

This task removes the live target for production websites from the Print Publications so that only a Preview and Live target are displayed during publishing. Prior to this there were two live targets.

This task is to test the new dedicated broker database and SDL discovery service to see if this fixes the Print Publications problem.

This task is to republish all the current content in the Preview Broker database for Print Publications.

This task creates a new target type and publication target specifically for Print Publications. When publishing to Live the print publications will have a new Live target that is only available for Publications.

New SDL Services were installed that are dedicated to the Print Publications website.

This completes the basic data architecture plan for storing all data related to the Media Library digital assets.

Azure contains hundreds of resources and without standardized naming practices it can be very difficult to locate resources. This task documented a new system we will use to name Azure objects in a way that we all understand and can easily group and locate resources.

This is a proof of concept to design a replication model so that our global indexes stay in sync.

Result: This work resulted in excellent results. We now have the ability to replicate a search index from one region to another in less than 30 minutes with strong potential for that to go down as fast as 2 minutes once we dedicate proper resources.

There are mulitple pricing models and ways to architect search indexes. This task is to select the most reliable and most cost effective model that will work with our search indexes.

Result: We will use three production indexes in North Central USA, East Coast(2) USA and Northern Europe and then a single instance for all the development environments. This provides the most cost effective solution by combining search indexes based on environment and not project. Previously we separated the indexes for T3 and YM which led to underutilized resources.

We rely heavily on search indexes to run most of our websites. This task is to explore the options to take our current search indexes which reside only in one part of the USA and distribute them globally so that in the event of the indexes going down in one region, we failover to another region. Microsoft Azure does not offer this type of functionality so we must create a setup that works for our content.

The resolution is that we can develop software that will allow our indexes to be globalized in a scalable manner that is fault tolerant when cloud hosting regions fail.