itHub user: andydesanctis (personal account,

no org)

Affected services (all show "Bad credentials"

on Branch connected to production):

| Project | Service

      | Service ID

  | Source repo

  |

|---------------------------|----------------

--------|------------------------------------

----|----------------------------------------

----|

| UpSkillr-Practice-Lab-SF |

practice-lab-sf-app |

bb7dae5e-9462-4e89-ba40-9295566d60d0 |

andydesanctis/UpSkillr-Practice-Lab-SF |

| MI-Practice-YRI |

UpSkillr-Practice-MI |

eeca0395-1a8c-4a2b-a203-3b223b0b0f11 |

andydesanctis/UpSkillr-Practice-MI |

| UpSkillr-Practice-AI-Demo | UpSkillr

Product Demo |

365de448-4988-4841-bf42-3800694cdcf4 |

andydesanctis/UpSkillr-Practice-AI-Demo |

Symptom: persistent "Bad credentials" on

Branch connected to production field on every

GitHub-sourced service. No webhook fires,

auto-deploy on push does not trigger.

Production is up (services still running on

prior successful deploys); only the

auto-deploy pipeline is broken.

What I've tried (all failed to clear the

error):

• Verified Railway GitHub App installed on my

GitHub account with access to all three

affected repos

• Fully uninstalled the Railway GitHub App

from GitHub via Settings → Applications →

Installed GitHub Apps → Uninstall, then

reinstalled via Railway's connect flow with

"Only select repositories" + all three repos

granted

• Confirmed via GitHub the new installation

has the correct repos in scope

• Per-service Disconnect → Reconnect on the

Railway dashboard for practice-lab-sf-app

(the disconnect appears to have cleared its

source field but reconnect still produces Bad

credentials)

• Other affected services

(UpSkillr-Practice-MI, UpSkillr Product Demo)

still show their repo binding but with the

Bad credentials error

A latest failed deploy for

practice-lab-sf-app: deployment ID

f1a08b6f-9ae4-4837-a7d6-17f8377907b6, FAILED

at 2026-05-23T12:39:09Z,

deployment_stopped=true. Confirms the

source-fetch stage is failing, not the build

or release process.

Diagnosis (per a similar resolved case in

your forums from 2026-04-29 by ppartlow on

ai-business-copilot): Railway's GitHub

integration state for my account has stale

internal state — likely cached

installation_id from the prior GitHub App

installation. Reconnect updates the visible

repo selection but doesn't refresh the

underlying credentials binding.

Note: there was a Railway platform

connectivity incident earlier today showing

"resolved" by the time I started

troubleshooting, but the timing correlates.

Mentioning in case it's relevant context for

diagnosing whether my account state was

caught in the incident.

Request: Please reset the GitHub integration

state at the account/installation level so

the cached installation_id is cleared across

ALL three affected services. After the reset,

next reconnect attempts on each service

should force a fresh handshake with GitHub.

Workaround in use: none (auto-deploy

blocked). Production services still running

on their prior successful deploys. `railway

up` from CLI would work for one-off deploys

but is impractical for normal development

flow.

Thank you.