Where did my projects go?
I had 9 or 10 projects on my neon account, its been a while since I created the last one or checked the dashboard, today I opened the dashboard and there's just 1 project there. The attached screenshot is a proof that I did have those projects. The DBs are working fine on the websites I used them in, which means they are not deleted, but can't access them in the dashboard.
Extremely slow vector nearest neighbour query (~30sec-2min) on cold database
I have roughly ~1 million 3072 dimension halfvecs in an embedding table, with a HNSW index.
When I perform a similarity-based query on the table, if the database is cold (scaled to zero for long enough for any caches to clear out), the query can take between 30 seconds to 2 minutes. I have verified that the index is being hit through running the query in the EXPLAIN ANALYSE view in the NeonDB console. The initial cold query is extremely slow. The subsequent query runs are very fast, but all the metrics in the EXPLAIN ANALYSE view (except time on the index scan) remain extremely similar.
The database cold start itself is not the problem, as to wake the database, I run a simple query that gets the version of an extension, which takes less than a second....
SOC2 Documents Request Denied But on Scale Plan?
Hi there we pay for a the scale tier; however I have been rejected from the SOC2 document portal. I tried to create a ticket in the web app, but it loads forever (sc attached). I have emailed feedback@neon.tech and still no response. really poor. Who can I talk to about this / how can I get these documents ASAP we are trying to get our SOC2 done!!!
Storage issue
Hello, I am currently experiencing a problem with storage, which indicates that I have exceeded my storage limit despite controlled usage.
Consumed CPU exceeded "allocated" CPU
Please refer to the screenshot, it seems certain queries are "exceeding" my configured CPU limit. Is this normal? Will my CU allotment be drained 4x as fast even though I set it to only use 0.25 vCPU and autoscaling is disabled?
High Compute Hour Usage in Free Tier
Hi Neon team/community! š
I'm seeing surprisingly high compute hour usage (75Ā hours this month for my branch) despite only doing solo, minor development testing. My active CU usage is consistently around 0.02, and my inactivity threshold is set to 5 mins.
My monitoring graph confirms this: the endpoint is mostly inactive, and the periods it is allocated show near-zero CPU usage (blue line). Even so, my usage has climbed by 7+ hours in the last 24 hours alone (68ā75Ā hours)....
Paid/Unpaid Support Issue & Refund Status
Issue: unable to contact Support due to an ongoing billing/refund issue. I'm stuck between paid and non-paid support and cannot get response from the Neon team. Can you please advise on how this can be resolved?
Details:
-I opened a support ticket while I was a paying customer, subject "[Neon Support] Re: Unable to downgrade Neon database for betterai-v1 - Vercel Support said only Neon support can fix the issue".
-The issue was resolved and was told explicitly by the Support person that I would get a refund. The last communication was 6 days ago from Neon support, there has been no refund....
DB goes down frequently - CONNECT_TIMEOUT errors
I'm using a serverless db for a nextjs vercel application. The db seems to be going down (scaling down) multiple times in a week - leading to 'CONNECT_TIMEOUT' issues on production - and I have to manually reboot the db to scale up. From what I understand, this is happening whenever allocated CPU compute units is not greater than used units.
I have already started a support ticket regarding this, but since it's taking some time for the Neon team to get back - I figured I'd start a thread to see if others have faced similar issue...
