Resized database app can't find volume

, ,
1

My postgres app recently stopped working because I hit the size limit, so I ran fly volumes extend. When I went to deploy I had to add an image, so I also added image = "postgres:16.2-bullseye" in the [build] section.

Now I’m seeing this and the database won’t start up.

16:48:08
Error: Database is uninitialized and superuser password is not specified.
16:48:08
       You must specify POSTGRES_PASSWORD to a non-empty value for the
16:48:08
       superuser. For example, "-e POSTGRES_PASSWORD=password" on "docker run".
16:48:08
       You may also use "POSTGRES_HOST_AUTH_METHOD=trust" to allow all
16:48:08
       connections without a password. This is *not* recommended.
16:48:08
       See PostgreSQL documentation about "trust":
16:48:08
       https://www.postgresql.org/docs/current/auth-trust.html

I don’t know how this can be true, this app’s been running for two years and I just had to resize it because there was too much data.

How do I troubleshoot this?

2

Hm… Extending the volume shouldn’t have required a redeploy…

How was this database originally created (two years ago)?

(The ones that you get automatically via the Fly.io platform wouldn’t carry a postgres: image.)

4

I don’t really remember. I think it’s an app just for postgres? It predates managed postgres.

This is my fly.toml from back then (the only addition since the original deployment is the [build] section).

app = 'app-name'
primary_region = 'ewr'

[build]
  image = "postgres:16.2-bullseye"

[env]
  FLY_SCALE_TO_ZERO = '1h'
  PRIMARY_REGION = 'ewr'

[[mounts]]
  source = 'pg_data'
  destination = '/data'

[[services]]
  protocol = 'tcp'
  internal_port = 5432
  auto_start_machines = true

[[services.ports]]
    port = 5432
    handlers = ['pg_tls']

  [services.concurrency]
    type = 'connections'
    hard_limit = 1000
    soft_limit = 1000

[[services]]
  protocol = 'tcp'
  internal_port = 5433
  auto_start_machines = true

[[services.ports]]
    port = 5433
    handlers = ['pg_tls']

  [services.concurrency]
    type = 'connections'
    hard_limit = 1000
    soft_limit = 1000

[checks]
  [checks.pg]
    port = 5500
    type = 'http'
    interval = '15s'
    timeout = '10s'
    path = '/flycheck/pg'

  [checks.role]
    port = 5500
    type = 'http'
    interval = '15s'
    timeout = '10s'
    path = '/flycheck/role'

  [checks.vm]
    port = 5500
    type = 'http'
    interval = '15s'
    timeout = '10s'
    path = '/flycheck/vm'

[[metrics]]
  port = 9187
  path = '/metrics'
5

Ok… That looks like PG Flex, in which case you would need an image of the form flyio/postgres-flex:<version>, where <version> would have to match what the Machine was configured with before the recent attempted redeploy, :sweat_smile:…

If you have a fly.toml, then you may have done a deploy years ago, in which case you might be able to just use fly releases --image to find the old image details in the history.

(Otherwise, it will require some detective work within the volume itself, etc.)

Aside: I would make a copy of the volume at this point, with fly vol fork, just in case there are mishaps…

6

You are an excellent person. Your advice was perfect, everything’s back up. Thank you very much!