DocumentAI fine-tuned custom extractor stuck in "undeploying" status for days

[ryanhaub]

I have two DocumentAI fine-tuned custom extractors stuck in "undeploying" status for days. They get deployed and undeployed on a schedule using Cloud Scheduler and Cloud Functions. That worked fine for months, and I haven't changed anything recently, so I'm not sure what happened.

I've tried manually undeploying and deploying them, but it tells me I can't do either because it's already "undeploying".

I see no way to forcefully stop the undeploy process. I see in the logs when the issue started, but no initial error was thrown. I only know it started then because I started getting the "400 ProcessorVersion state cannot be changed to 'UNDEPLOYING' since it is 'UNDEPLOYING'." error each time the scheduler would run.

Any ideas on how I can forcefully stop the undeploy process or otherwise reset it. I don't want to delete the processor and start over because I have a ton of labeled documents and I've already paid for the fine-tuning of the model to get it just right. I don't want to have to redo all of that.

[cassandramae · Accepted solution]

Hi @ryanhaub,

Welcome to Google Cloud Community!

Typically, this issue is resolved by undeploying and then redeploying. However, this solution didn't seem to work in your case. With this, to isolate any browser-related issues, please try undeploying using a different supported browser. You may also check the browser-related troubleshooting here which might help.

If the issue persists, I suggest reaching out to Google Cloud Support for further assistance on how to undeploy your custom extractor on their end.

Was this helpful? If so, please accept this answer as “Solution”. If you need additional assistance, reply here within 2 business days and I’ll be happy to help.

[ryanhaub]

I ended up reaching out to Cloud Support and they were able to fix it for me. I wish there was a way to diagnose and resolve on my own, but at least the problem was resolved.