Google Cloud Vertex AI Endpoint deployment is extremely slow

Question

I have built a model using Google Cloud Vertex AI and uploaded to Model Registry. I then try to create an Endpoint with it but I found that it is extremely slow - takes more than 30-40mins. And sometimes it just stick as deploying... Is this normal? Is there any way to speed this up? Or at least is there anywhere I can see any possible error?

Grogan

Having the same issue, will like to know if you managed to figure out why – Jonejonell

Even I am facing same problem. – Belike

I am wondering about this too. – Rauch

about a year later, still seeing this problem. – Auctioneer

same issue, while waiting endlessly , got completion email – Scorper

Having the same issue in May 2024, and a link to logs weirdly comes in an email instead of being displayed in the vertex console. – Ulick

Answer 1

I had similar issue and after few failed attempts with instance based Notebook option which is default oe, I switched to managed notebook and it did work phew. Managed notebook is the 2nd Tab next to Instance. After couple of options, it stopped creating new instances for me.

Hope this helps

Scorper