We're seeing private-network instability in us-west2 affecting two environments in project RTA (fa5e61ff-3d5f-4920-97f2-a5d0bff33dc4). Both app services connect to Postgres over postgres.railway.internal.

We haven't touched the project / made any changes in a few months. This came out of nowhere.

Network flow logs show TCP packets marked invalid between our app and rta-db services.

production (27bc7bbc-…): app fails with getaddrinfo … errno 3, code "ESERVFAIL" resolving postgres.railway.internal: DNS returning SERVFAIL. Persists after redeploying app.

staging (4cb50a07-…): DNS resolves, but DB connections drop mid-stream, causing intermittent failed writes (our OAuth state rows fail to persist).

No ESERVFAIL, but consistent with dropped established connections. Ruled out on our side

rta-db is healthy in both envs (clean startup, normal checkpoints, no errors).

DATABASE_URL correctly uses the internal host; no config changes.

App redeploy did not clear it.

This points to platform-level private-networking packet loss / internal-DNS failure in us-west2, not our configuration. Can you check private-network and internal-DNS health for this project/region?