Can’t connect to Yandex Cloud Postgres DB

Can’t connect to Yandex Cloud Postgres DB from Vercel – getaddrinfo ENOTFOUND


Current versus Expected behavior

  • Current behavior:
    When trying to connect to a Yandex Cloud Postgres database from my Vercel project, I receive the error:
    getaddrinfo ENOTFOUND rc1a-hhvlknmp5l43fk7n.mdb.yandexcloud.net
    This results in a 500 error for all API calls that access the database.
    DNS resolution for the Yandex host fails inside Vercel’s environment.
  • Expected behavior:
    The Vercel backend should be able to resolve the Yandex Cloud Postgres hostname and connect successfully, as it works in my colleague’s project with the same setup.

Code, configuration, and steps that reproduce this issue

  • Code / Config
    I am using standard Postgres connection in Node.js / Next.js API routes (see code snippets below, if required).
    My environment variables:

ini

КопироватьРедактировать

POSTGRES_HOST=rc1a-hhvlknmp5l43fk7n.mdb.yandexcloud.net
POSTGRES_PORT=5432
POSTGRES_DB=FERMA
POSTGRES_USER=lolmeteor
POSTGRES_PASSWORD=********

No custom networking or DNS code.
The DB cluster is in Yandex Cloud, open for external connections (public host).

  • Steps to reproduce:
    1. Deploy backend project on Vercel.
    2. Configure environment variables for Yandex Cloud DB (as above).
    3. Deploy and call any API endpoint that queries the database.
    4. Observe error 500 and log message:
      [Error: getaddrinfo ENOTFOUND rc1a-hhvlknmp5l43fk7n.mdb.yandexcloud.net]

Project information (URL, framework, environment, project settings)

  • Vercel project URL:
    https://v0-magaz-supabase.vercel.app
    (also tried on test deployments)
  • Framework:
    Next.js / Node.js API routes
  • Environment:
    Vercel Serverless Functions, default region
  • Project settings:
    No custom DNS, using standard Vercel/Next.js deployment

Additional notes:

  • My colleague has an identical project on Vercel, using the same Yandex Cloud DB cluster and hostname, and everything works fine for him.
  • I already checked firewall, whitelists, credentials, public accessibility, and DNS from my local machine (all OK).
  • I spent $20+ troubleshooting this without result.
  • Please have a real human support engineer review my issue (not an automated reply).

Logs:
Jun 04 14:35:51.89
GET
200
v0-magaz-supabase.vercel.app
/product/t-shirt-1747811562277
Jun 04 14:35:51.89
GET
200
v0-magaz-supabase.vercel.app
/product/t-shirt-1748420095791
Jun 04 14:35:51.89
GET
200
v0-magaz-supabase.vercel.app
/product/t-shirt-1747811660556
Jun 04 14:35:19.45
POST
500
v0-magaz-supabase.vercel.app
/api/yandex-db/users/create
3
Error creating user: [Error: getaddrinfo ENOTFOUND rc1a-hhvlknmp5l43fk7n.mdb.yandexcloud.net] { errno: -3008, code: ‘ENOTFOUND’, syscall: ‘getaddrinfo’, hostname: ‘rc1a-hhvlknmp5l43fk7n.mdb.yandexcloud.net’ }
Jun 04 14:35:18.59
GET
200
v0-magaz-supabase.vercel.app
/profile
Jun 04 14:35:18.59
GET
200
v0-magaz-supabase.vercel.app
/cart
Jun 04 14:35:18.46
POST
500
v0-magaz-supabase.vercel.app
/api/yandex-db/users/by-telegram-id
3
Error in users/by-telegram-id: [Error: getaddrinfo ENOTFOUND rc1a-hhvlknmp5l43fk7n.mdb.yandexcloud.net] { errno: -3008, code: ‘ENOTFOUND’, syscall: ‘getaddrinfo’, hostname: ‘rc1a-hhvlknmp5l43fk7n.mdb.yandexcloud.net’ }
Jun 04 14:35:15.49
GET
200
v0-magaz-supabase.vercel.app
/
Jun 04 14:33:14.13
GET
200
v0-magaz-supabase-qmppzyyxo-lolmeteor-gmailcoms-projects.vercel.app
/cart
Jun 04 14:33:14.13
GET
200
v0-magaz-supabase-qmppzyyxo-lolmeteor-gmailcoms-projects.vercel.app
/profile
Jun 04 14:33:12.36
GET
200
v0-magaz-supabase-qmppzyyxo-lolmeteor-gmailcoms-projects.vercel.app
/
Jun 04 14:33:12.34
GET
200
v0-magaz-supabase-qmppzyyxo-lolmeteor-gmailcoms-projects.vercel.app
/
Creating user with Telegram ID: 5137953484
Ошибка выполнения запроса к Yandex DB: [Error: getaddrinfo ENOTFOUND rc1a-hhvlknmp5l43fk7n.mdb.yandexcloud.net] {
errno: -3008,
code: ‘ENOTFOUND’,
syscall: ‘getaddrinfo’,
hostname: ‘rc1a-hhvlknmp5l43fk7n.mdb.yandexcloud.net
}
Error creating user: [Error: getaddrinfo ENOTFOUND rc1a-hhvlknmp5l43fk7n.mdb.yandexcloud.net] {
errno: -3008,
code: ‘ENOTFOUND’,
syscall: ‘getaddrinfo’,
hostname: ‘rc1a-hhvlknmp5l43fk7n.mdb.yandexcloud.net
}
Fetching user with Telegram ID: 5137953484
Ошибка выполнения запроса к Yandex DB: [Error: getaddrinfo ENOTFOUND rc1a-hhvlknmp5l43fk7n.mdb.yandexcloud.net] {
errno: -3008,
code: ‘ENOTFOUND’,
syscall: ‘getaddrinfo’,
hostname: ‘rc1a-hhvlknmp5l43fk7n.mdb.yandexcloud.net
}
Error in users/by-telegram-id: [Error: getaddrinfo ENOTFOUND rc1a-hhvlknmp5l43fk7n.mdb.yandexcloud.net] {
errno: -3008,
code: ‘ENOTFOUND’,
syscall: ‘getaddrinfo’,
hostname: ‘rc1a-hhvlknmp5l43fk7n.mdb.yandexcloud.net
}


End of template.

This seems like a valid error. From what I can see, rc1a-hhvlknmp5l43fk7n.mdb.yandexcloud.net genuinely does not exist.

1 Like

Hi Amy,

Thank you for the reply.

I have to clarify:
rc1a-hhvlkhmp5l43fk7n.mdb.yandexcloud.net is a valid FQDN provided by Yandex Cloud.
You can see it in my Yandex Cloud console screenshot attached below. This is the official hostname for my Postgres cluster. My colleague is running an identical Vercel project using exactly this hostname, and his project connects with no issues. We are using the same credentials, cluster, and host.

To be 100% clear:

  • The hostname is correct, public, and resolves in most environments.
  • This is not a typo, and not a local network restriction — see the screenshot from Yandex Cloud and environment variable list from my colleague’s working Vercel project (also attached).
  • When I run nslookup or ping on this FQDN from outside of Vercel (locally or from other cloud services), DNS resolves fine.
  • Only my Vercel project gets the ENOTFOUND error; my colleague’s Vercel deployment (same cluster, same code, same env vars) connects successfully.

Screenshots attached:

  1. [Yandex Cloud console with FQDN highlighted]
  2. [Environment variable settings from my colleague’s Vercel dashboard]

Request:
Can you please escalate this to an engineer familiar with your internal DNS or networking stack? There is clearly something inconsistent on Vercel’s end, since the same setup works for one account/project and fails for another.

Let me know if you need any more technical details, additional logs, or live DNS output.
Thank you.

It doesn’t seem to be just my computer that’s unable to locate that address -https://www.nslookup.io/domains/rc1a-hhvlknmp5l43fk7n.mdb.yandexcloud.net/dns-records/

I recommend a few things in this case…

  1. You need to let your colleague know that you exposed their passwords and keys on a public website. The passwords and keys need to be changed now. I deleted the screenshot for you, but the info has already been exposed
  2. Double check that the address is correct with no typos
  3. Work with your colleague to determine how their setup differs from yours