Skip to content
Postgres removed as “Compromised service” on 2 projects (easy-order + fms) — need restore
navedk0201
OP

10 days ago

Hi — need help restoring Postgres. Same issue on two production projects.

Error on deploy:

“Compromised service detected… may violate Terms of Service or Acceptable Use Policy, so it was removed. Repeated attempts may result in a permanent ban.”

Workspace: navedk0201's Projects

Account: [email protected]

Region: asia-southeast1-eqsg3a (Singapore)

  1. easy-order

Project ID: 11c50b6a-9d74-4306-8ace-0e6921ac8287

Env: production (a2dfe041-1f4a-4339-b866-c359082e69c4)

Postgres service: 43cb16f1-3828-4dfa-839d-3582c8472284

Volume: postgres-volume (1a915ce1-4639-410b-89e5-ce5b079b2ea4)

Status now: deploymentId null (not running)

Compromised REMOVED deploys:

  • 33f852fd-583e-40d2-b7ca-a4244777b953 (2026-11-03)
  • caea363c-ba1b-4685-9971-0d9fd0c13dc1 (2026-10-29)
  • 392fda3c-80b6-469a-8846-29a21bdbc5ea (2026-08-29)
  1. fms

Project ID: 375ed4e0-0931-4c11-9d4f-342685597c6b

Env: production (7d2a2184-2370-4bc3-acd8-3c2f5fde54e6)

Postgres service: 3334e18a-eecd-461f-ac84-f7c2887824fd

Status now: FAILED / stopped

Compromised REMOVED deploys:

  • ed324f02-cab0-4c44-bcbc-33e666d8690b (2026-11-08)
  • 779f1983-cdde-4d00-bf56-e86c30be8b0b (2026-11-07)
  • 1e375d88-97bb-4379-9c82-9b8537010074 (2026-11-03)
  • 7209fa74-dec9-418f-864e-1fd1ad8cb887 (2026-10-31)
  • 7b437c56-e55e-4843-afb0-857a793c5ba4 (2026-08-29)

Both are legitimate apps (EasyOrder + FMS). Backend services are down because DB is gone. We’ve stopped redeploying after the ban warning.

Please:

  1. Confirm why these were flagged

  2. Clear the flag if false positive

  3. Allow redeploy + remount existing volumes to recover data

Thanks.

Solved

4 Replies

Status changed to Awaiting Railway Response Railway • 10 days ago


10 days ago

Your data is not lost. The Postgres volumes in production on both easy-order and fms are still attached to their Postgres services and still hold their data. The flag removed the running deployments, not the volumes, so nothing needs to be remounted.

The "Compromised service detected" removal applies only to those Postgres services. It is not a restriction on your workspace or account. Each detection removes the deployment again, so a plain redeploy will be removed within minutes until the service is secured.

Before you redeploy each Postgres service, remove or lock down any public access to the database. Then rotate the database password and any credentials that can be reached from it, update the services that connect to it, and redeploy. Treat the current data as untrusted and check it for anything you did not create before you rely on it.


Status changed to Awaiting User Response Railway • 10 days ago


brody

Your data is not lost. The Postgres volumes in production on both easy-order and fms are still attached to their Postgres services and still hold their data. The flag removed the running deployments, not the volumes, so nothing needs to be remounted. The "Compromised service detected" removal applies only to those Postgres services. It is not a restriction on your workspace or account. Each detection removes the deployment again, so a plain redeploy will be removed within minutes until the service is secured. Before you redeploy each Postgres service, remove or lock down any public access to the database. Then rotate the database password and any credentials that can be reached from it, update the services that connect to it, and redeploy. Treat the current data as untrusted and check it for anything you did not create before you rely on it.

navedk0201
OP

10 days ago

Hi Brody,

Thanks for confirming that both production PostgreSQL volumes and their data are intact.

However, our main concern is why Railway is repeatedly detecting both PostgreSQL services as compromised and automatically removing their deployments.

This has happened multiple times since August 29, 2026, across two independent production projects (easy-order and fms).

Could you please help us identify:

  1. What exact suspicious activity or security signature triggered these detections?
  2. Was an unauthorized process, external connection, malicious SQL command, or known vulnerability detected?
  3. Can you share the relevant timestamps, process details, source IPs (where available), and security logs?
  4. Is the issue related to publicly exposed PostgreSQL endpoints, compromised credentials, or the deployed PostgreSQL Docker images?
  5. How can we confirm that the databases are secure before redeploying and avoid repeated automatic removals?

We understand the recommendation to restrict public access and rotate credentials, and we will take these precautions.

However, we need the specific detection reason and root cause, rather than repeatedly redeploying without knowing what triggered the security flag.

As both production applications are affected, please escalate this to the Railway Security/Infrastructure team for investigation.

Thanks.


Status changed to Awaiting Railway Response Railway • 10 days ago


10 days ago

We don't share the specific detection signals, rule details, or security logs behind a compromised-service flag, because that information would help someone evade detection. What we can tell you is that this enforcement is triggered by activity coming from the service itself, it applies only to the flagged Postgres services, and it is not a restriction on your workspace or account.

The most useful thing to look at is how each database can be reached from the internet. Both Postgres services have a public connection URL variable set, and the fms production Postgres service also has a public Railway domain pointed at the database port. Remove any public networking on these services (public domains and TCP proxies) so they are reachable only over private networking from your own services. If something outside Railway genuinely needs direct access, keep that path closed until the database is secured.

After that, change the Postgres password on each service, update every service that connects to it, and rotate any other credentials that were stored in or reachable from those databases. Then redeploy each Postgres service, which starts it from a fresh copy of the image against your existing volume.

Before relying on the data, review it for roles, extensions, functions, scheduled jobs, or tables you did not create, since anything added by an unauthorized party lives in the volume and would survive a redeploy. If the service is flagged again after these steps, that indicates something in the database itself is still active and needs to be removed.


Status changed to Awaiting User Response Railway • 10 days ago


brody

We don't share the specific detection signals, rule details, or security logs behind a compromised-service flag, because that information would help someone evade detection. What we can tell you is that this enforcement is triggered by activity coming from the service itself, it applies only to the flagged Postgres services, and it is not a restriction on your workspace or account. The most useful thing to look at is how each database can be reached from the internet. Both Postgres services have a public connection URL variable set, and the fms production Postgres service also has a public Railway domain pointed at the database port. Remove any public networking on these services (public domains and TCP proxies) so they are reachable only over private networking from your own services. If something outside Railway genuinely needs direct access, keep that path closed until the database is secured. After that, change the Postgres password on each service, update every service that connects to it, and rotate any other credentials that were stored in or reachable from those databases. Then redeploy each Postgres service, which starts it from a fresh copy of the image against your existing volume. Before relying on the data, review it for roles, extensions, functions, scheduled jobs, or tables you did not create, since anything added by an unauthorized party lives in the volume and would survive a redeploy. If the service is flagged again after these steps, that indicates something in the database itself is still active and needs to be removed.

navedk0201
OP

10 days ago

Hi Brody,

Thank you for your support and guidance.

I have removed the public networking access as recommended. We will continue monitoring both PostgreSQL services and take the necessary security precautions.

If the issue occurs again, I will report it here for further investigation and assistance.

Thanks again for your support.


Status changed to Awaiting Railway Response Railway • 10 days ago


Status changed to Solved Railway • 10 days ago


Welcome!

Sign in to your Railway account to join the conversation.

Loading...