Continuing Our Content Catalog Quest

Sorry, this one is long. Putting it out there to hopefully help someone else. If you don’t want to read the whole thing, here’s the BLUF:

• Content Catalog has poor tools for batch management and they are spread across different sites
• I am reluctant to trust that the Content Catalog database is as clean as it thinks it is, due to discrepancies between the UI and API
• The reliance on browser cache is awful and feels like a hack to me

THE PLAN

As I mentioned in a previous post, my plan for Content Catalog is to burn it to the ground so I can build it back up.

This has been the definition of “easier said than done.”

My original plan was to go to the page on Forma admin where you turn ON Content Catalog, and switch it to OFF.

Well, guess what you can’t do? Once you turn ON Content Catalog, you are committed. You can hide it from users. You can (in theory) empty it out, but you can never turn it off.

So, I had several things to deal with:

• Removing user access
• Deleting the files
• Clearing the tags
• Removing the collections

Part of the challenge here is that managing Content Catalog is not a one-stop shop: Users and collections are managed in the Forma admin pages. Files and tags are tweaked in the Content Catalog pages. Not really difficult, but definitely makes one scratch their head and ask “why?”

LET’S START PURGING

Removing users was a piece of cake. Go to the Forma admin pages, delete the groups, clean up others and voila! User access removed. Leave a small handful of folks as admins just in case.

Then came the task of deleting the stuff. With zero batch management tools. In theory, you can select a bunch of things and click a button, but the web UI is very poor at tackling activity on a large number of items at once. It gets confused, seems to time out, and a browser is no place for that kind of work. So essentially, no batch management.

We need to get rid of files, collections, and tags. You can’t remove tags or collections if they have stuff in them, so the first step is to delete the files. I am dealing with thousands of files. Many thousands. And did I mention there are no batch management tools? Every web search says to do a search in Content Catalog, then hit the check all button and delete them that way. So I tried that in chunks, with unknown success. Some rows would disappear and then I got errors like “Failed to write delta sync record.” and “Conflict resolver rejects mutation.”

Clear as mud.

Those are, apparently, database errors. My guess is that some of my manual deletion worked but it wasn’t reporting back properly?

I kept trying to brute force the deletions in batches until I got mad. Then I just let it sit over the weekend.

LET’S JUST GIVE IT THE WEEKEND AND SEE

Tuesday morning, hopped back in… still over 20,000 items. The web page was reporting that I basically did nothing.

So it was time for some vibe coding with Claude. The current v2 beta of the Content Catalog API says that it can do deletions. Sweet. First though, Claude and I did an inventory. Expecting about 20k items, imagine my surprise when the API reported 179 items. And digging into them, several were for Revit 2021, which is no longer supported.

Then I looked at my Content Catalog in a new Incognito mode browser. The same 179 objects.

THE API FOR THE WIN… MAYBE

It was the browser cache giving me false results. Almost as much as performance issues, the reliance on the browser as delivery and index method really turns me off. A refresh didn’t reset the cached index. A hard refresh doesn’t either. You have to do a big fat cache clear. This feels unsustainable and short sighted to me… and certainly adds to the list of why CC does not support batch edits/updates/etc. well.

So, some more deletions using the API and little Python scripts Claude put together for me (with caps on number of actions from the API and, again, not something for batch updates). Then a big fat browser clear. Finally in the browser, I had 0 items, 0 collections… and 725 tags?

The API agreed. 725 tags. The API also said that the tags were tied to some piece of content so they couldn’t be deleted. But there is no more content… right? Oddly, I could manually delete those tags through the website, no error.

YET ANOTHER TOOL

I thought I could get Claude to click the trash can plus two confirmation dialog on the website to delete my Tags, but it said in no uncertain terms that while it could accomplish the “mouse click’ actions, it was not allowed to delete. Anything. And I keep getting told that AI is going to kill all humans.

Some quick searching got me to Power Automate. That was simple to configure and could loop through my Tags and get rid of them.

But, why another tool? Why not a simple mass editing feature right in Content Catalog itself?

A WARNING

So, I document all of this as a warning if you need to do any kind of mass updates to your Content Catalog library. Right now, it sucks. The inconsistency with the API and UI reporting, the abysmal performance in batch deletions, the lack of an index refresh without doing something in my browser… But, at the en off the day these are mostly a me problem. The end users should rarely see these issues, so I have to take that into account when evaluating if this is the right solution for the firm.

But my #1 concern is my lack of trust. The conflicting numbers, the slow updates, the weird errors, the secret content, the tags tied to objects when there are none… I am extremely dubious that my clean database is actually clean, and I have no way to know if these ghosts will impact performance. I do have a ticket open with Autodesk to discuss… I’ll report on how that goes when it happens.