Before you report this, we’d appreciate if you can search this forum to see if this issue has already been reported.
If you have done so, please check this box.
Describe the Bug
I am trying to index a site (Cloudflare Provider | Pulumi Registry) and every time it gets to ~190 pages and then just returns “Indexing failed”. Retrying a number of times does not resolve the issue.
When then trying to chat with what docs it does have in the index, it fails with “Non-turbopuffer vector databases are no longer supported”
Current Cursor Version (Menu → About Cursor → Copy)
Version 0.44.9
Request ID: e6c15d4c-16d7-48f7-8019-3605629e94d8
Additional Information
No other additional information I can think of. The only thing I notice is I think it’s downloading some existing embeddings from somewhere when I first add the site, and that’s what’s being submitted because the index fails.
A shame to see this keeps reoccuring; here’s another thread I raised about all the same things back in June:
The price issues are particularly stinging, because despite the price of all the LLMs coming down, the “pro” requests are still fixed at their 500 calls, and even when a call fails inside of Cursor, you still get billed, anyway ¯_(ツ)_/¯
It’s easy to slam AI IDEs - people are trying to do something remarkable and something we never had before. That being said, the constant issues and bugs which sometimes block critical AI features working for “random periods of time” with almost no communication (and certainly no refunds or service credits for the continuous problems) could be solved with better transparency and communication, still.
Well, I’m not sure if that’s the solution, but I deleted the index and all docs once. Then reset everything again and currently it works. However, I can’t say yet whether it will stay this way.
To me it looks like it’s not working properly, or at least I can’t get it to work properly still, although don’t know how to debug it any further than the following:
Have a failed doc indexing task (or appears to be any task?), click “reindex”
Reindex immediately reports “Indexed”
Result:
The documents do not appear to be indexed.
I checked all the various “Outputs” I could for Cursor, but couldn’t find anything. Do please let me know if there’s any further way I can help.
Yeah that’s right, its not crawling through the docs anymore, its busted… don’t know when it will be fixed… It only goes to the first introduction page and then that’s it, no further @curosr