Retry deployment after 10 seconds 3 Times

[0tuedon]

From my research, the error mainly happens with Railway (sometimes the authentication skips), and then it fails. I’m adding a retry mechanism that will try 3 times. I believe it will work because when I manually restart the workflow, it usually passes the second time. If this doesn’t work, we can easily undo this change.

cc @satsie , @jrakibi

[satsie]

Sounds good to me! Let's give it a go and see if the failures go away 👍

[jrakibi]

cross posting this comment here:

I think the the current script will exit with success even if all three attempts fail (because the last command executed is either sleep or echo both of which return a success status). This means the job might incorrectly report success even if the deployment fails.



We can add a check like this at the end with exit 1 command


- name: Deploy
  run: |  # This script retries the deployment 3 times before failing
    n=0
    until [ "$n" -ge 3 ]
    do
      railway up --service public-inbox && break
      n=$((n+1))
      echo "Retrying deployment... attempt $n"
      sleep 15  # Increased delay between retries
    done
    
    # Check if all retries failed
    if [ "$n" -eq 3 ]; then
      echo "All deployment attempts failed"
      exit 1 
    fi
  env:
    RAILWAY_TOKEN: ${{ secrets.RAILWAY_TOKEN }}

We can do this later. Let's first see what happens with the current changes merged and check the logs, even if it reports a success status

[satsie]

oh my! good catch @jrakibi!!! This should get addressed in a follow up PR.