Narven

Narven

Hi,

For the last 2 days, I’ve be trying to implement Triplex to add multi-tenancy to my application, and I use transactions a lot, usually with Ecto.Multi.

From the documentation it looks like with triplex I need to use:

    Triplex.create_schema(slug, Project.Repo, fn(tenant, repo) ->
      Repo.transaction(fn ->
        mig = Triplex.migrate(tenant, repo)
        IO.inspect(mig, label: "MIGRATE")

        // Add any ecto operation here...

        tenant
      end)
    end)

My problem is that Triplex.migrate(tenant, repo) is always, always returning:

 {:error,
 "ERROR 42P01 (undefined_table) relation \"prj_c24.schema_migrations\" does not exist\n\n    query: INSERT INTO \"prj_c24\".\"schema_migrations\" (\"version\",\"inserted_at\") VALUES ($1,$2)"}

For me this is an extreme confusing error. It tells nothing about what is happening, (at least for me).

My only migration is this:

defmodule Project.Repo.Migrations.CreateUsersTables do
  use Ecto.Migration

  def change do
    # execute "CREATE EXTENSION IF NOT EXISTS citext", ""

    create table(:users, primary_key: false) do
      add(:id, :uuid, primary_key: true, null: false)
      # add(:email, :citext, null: false)
      add(:hashed_password, :string, null: false)
      add(:first_name, :string, null: false)
      add(:last_name, :string, null: false)
      add(:is_enabled, :boolean, null: false, default: true)
      add(:dob, :date)
      add(:photo, :string, default: "/images/empty.png")
      add(:confirmed_at, :naive_datetime)
      timestamps()
    end

    # create unique_index(:users, [:email])
  end
end

Any help on that error? thanks

PS: The prefix is being created in the database. But migrations are not running.

❯ elixir --version
Erlang/OTP 24 [erts-12.1.5] [source] [64-bit] [smp:16:16] [ds:16:16:10] [async-threads:1] [jit] [dtrace]
Elixir 1.13.0 (compiled with Erlang/OTP 24) 
{:triplex, "~> 1.3.0-rc.1"}

Showing Posts 1 to 10

al2o3cr

al2o3cr

The error is complaining because the schema_migrations table isn’t present in the tenant schema.

That doesn’t make a ton of sense - Triplex.migrate calls Ecto.Migrator.run which ensures that table exists :thinking:

Is there anything useful in the Elixir or DB logs?

Narven

Narven OP

Yeah.. I was just looking at that. it looks like for some reason that “schema_migrations” is not being created on the tenants side.

Could it be because i have both “migrations” and “tenant_migrations” inside of “repo” folder ?

I was trying to see if there was a way to prevent the rollback. so that I could see the last state.

Narven

Narven OP

From the logs:

[debug] QUERY OK db=3.1ms queue=2.0ms idle=1959.2ms
CREATE SCHEMA "vsk_bn" []
[debug] QUERY OK db=1.2ms idle=967.2ms
begin []
TENANT: "bn"
REPO: Project.Repo
[info] == Running 20211010180955 Project.Repo.Migrations.CreateUsersTables.change/0 forward
[info] execute "CREATE EXTENSION IF NOT EXISTS citext"
[info] extension "citext" already exists, skipping
[info] create table vsk_bn.users
[info] create index vsk_bn.users_username_index
[info] == Migrated 20211010180955 in 0.0s
MIGRATE: {:error,
 "ERROR 42P01 (undefined_table) relation \"vsk_bn.schema_migrations\" does not exist\n\n    query: INSERT INTO \"vsk_bn\".\"schema_migrations\" (\"version\",\"inserted_at\") VALUES ($1,$2)"}
al2o3cr

al2o3cr

Getting to the == Running 2021etcetcetc bit suggests that the code believes it has created the schema_migrations table. :thinking:

One thought: this pattern looks like the one mentioned in the Triplex docs, but I wonder if that actually works. Maybe the schema_migrations table is being inserted in a separate transaction somehow; that would produce the observed behavior:

  • start transaction with Repo.transaction in your posted code
  • Triplex.migrate calls into Ecto.Migrator, which starts a transaction in another process (the migration lock holder?)
  • other transaction commits successfully, creating the table - but that isn’t visible to the first transaction that was already opened
  • Ecto.Migrator tries to read from schema_migrations and fails
Narven

Narven OP

One thing I dont understad is that if I go and run iex -S mix phx.server

and on iex i run:

iex(4)> Triplex.create("we")
[debug] QUERY OK db=1.6ms decode=1.5ms queue=1.4ms idle=928.7ms
CREATE SCHEMA "vsk_we" []
[info] == Running 20211010180955 Project.Repo.Migrations.CreateUsersTables.change/0 forward
[info] execute "CREATE EXTENSION IF NOT EXISTS citext"
[info] create table vsk_we.users
[info] create index vsk_we.users_username_index
[info] == Migrated 20211010180955 in 0.0s
{:ok, "we"}
iex(5)>

It works flawless.

If I run Triplex.create("foo") directly on the code… Does do nothing.

Because for some weird reason according to the docs we have to run create_schema. But on IEX just running create works perfectly.

Narven

Narven OP

If I do something simple like:

    slug = "foo"
    mig = Triplex.create(slug, Project.Repo)
    IO.inspect(mig, label: "MIG")

    Repo.insert(%User{
      first_name: attrs["first_name"],
      last_name: attrs["last_name"],
      username: attrs["username"],
      email: attrs["email"],
      hashed_password: Bcrypt.hash_pwd_salt(password)
    }, prefix: Triplex.to_prefix(slug))

it works. I think that the issue is only when they are running inside a migration.

al2o3cr

al2o3cr

Yeah, that’s what makes me suspicious about the Repo.transaction in the middle - Triplex.create is defined in a way that’s equivalent to:

Triplex.create_schema(slug, Project.Repo, fn(tenant, repo) ->
  Triplex.migrate(tenant, repo)
end)

Re: your last message about running inside a migration - if this code is inside a migration’s def up, then there’s already a transaction wrapped around the whole thing. You may need to disable that with @disable_ddl_transaction true.

Narven

Narven OP

It looks like adding @disable_ddl_transaction true to the migration and doing like this:

Triplex.create_schema(slug, Project.Repo, fn(tenant, repo) ->
      repo.transaction(fn ->

        mig = Triplex.migrate(tenant, repo)

        IO.inspect(mig, label: "~> Versions")
        repo.insert!(%User{
          first_name: attrs["first_name"],
          last_name: attrs["last_name"],
          email: attrs["email"],
          username: Timex.now |> to_string,
          hashed_password: Bcrypt.hash_pwd_salt(password)
        }, prefix: Triplex.to_prefix(slug))

        IO.inspect(tenant, label: "~> TENANT")
        tenant
      end)
    end)

returns:

[debug] QUERY OK db=1.8ms queue=1.4ms idle=1556.8ms
CREATE SCHEMA "vsk_e3" []
[debug] QUERY OK db=0.9ms idle=1560.1ms
begin []
[info] == Running 20211010180955 Project.Repo.Migrations.CreateUsersTables.change/0 forward
[info] execute "CREATE EXTENSION IF NOT EXISTS citext"
[info] extension "citext" already exists, skipping
[info] create table vsk_e3.users
[info] create index vsk_e3.users_username_index
[info] == Migrated 20211010180955 in 0.0s
~> Versions: {:error,
 "ERROR 42P01 (undefined_table) relation \"vsk_e3.schema_migrations\" does not exist\n\n    query: INSERT INTO \"vsk_e3\".\"schema_migrations\" (\"version\",\"inserted_at\") VALUES ($1,$2)"}
[debug] QUERY OK db=1.6ms
INSERT INTO "vsk_e3"."users" ("email","first_name","hashed_password","is_enabled","last_name","username","id","inserted_at","updated_at") VALUES ($1,$2,$3,$4,$5,$6,$7,$8,$9) ["e3@gmail.com", "e3", "$2b$12$c2/QHSJVJz3r9m6IzqZJI.ZDJWQnTYIIZrYW1EjJdUtcXBLCpUFUm", true, "e3", "2021-12-08 22:09:58.266144Z", <<92, 43, 112, 180, 231, 134, 79, 54, 175, 253, 215, 250, 153, 244, 217, 123>>, ~N[2021-12-08 22:09:58], ~N[2021-12-08 22:09:58]]
~> TENANT: "e3"
[debug] QUERY OK db=2.6ms
commit []

Still shows the error, which technally is an error, but in this case I dont think it should show has an error.

~> {:error,
 "ERROR 42P01 (undefined_table) relation \"vsk_e3.schema_migrations\" does not exist\n\n    query: INSERT INTO \"vsk_e3\".\"schema_migrations\" (\"version\",\"inserted_at\") VALUES ($1,$2)"}

I think this is more like… I cannot find schema_migrations… so I’m going to create it. Or maybe not.

But for now it looks like is working… I will add tomorrow all the extra removed migrations and add the rest of the transactions to see how it goes.

Thanks @al2o3cr

Narven

Narven OP

After more testing… this lib does not work at all.
This working if there is only one migration, after that for some reason it throws the error

 {:error,
 "ERROR 42P01 (undefined_table) relation \"vsk_p1.schema_migrations\" does not exist\n\n    query: INSERT INTO \"vsk_p1\".\"schema_migrations\" (\"version\",\"inserted_at\") VALUES ($1,$2)"}

and no other migration get’s run:

i’ve currently have 2 migrations and this is what happens:

[info] == Running 20211011183604 Project.Repo.Migrations.CreateSchools.change/0 forward
[info] create table vsk_p1.schools
[info] create index vsk_p1.schools_slug_index
[info] == Migrated 20211011183604 in 0.0s
§ TENANT: "p1"
§ REPO: Project.Repo
§ EXCEPTION: %Postgrex.Error{
  connection_id: 925,
  message: nil,
  postgres: %{
    code: :undefined_table,
    file: "parse_relation.c",
    line: "1139",
    message: "relation \"vsk_p1.schema_migrations\" does not exist",
    pg_code: "42P01",
    position: "13",
    routine: "parserOpenTable",
    severity: "ERROR",
    unknown: "ERROR"
  },
  query: "INSERT INTO \"vsk_p1\".\"schema_migrations\" (\"version\",\"inserted_at\") VALUES ($1,$2)"
}
~> MIGRATIONS: {:error,
 "ERROR 42P01 (undefined_table) relation \"vsk_p1.schema_migrations\" does not exist\n\n    query: INSERT INTO \"vsk_p1\".\"schema_migrations\" (\"version\",\"inserted_at\") VALUES ($1,$2)"}

PS: added some extra debug lines to the actual library §

riebeekn

riebeekn

I think the inner transaction is the issue. As mentionned in the Triplex docs the way migrations run was changed in Ecto 3. Since that change I haven’t ever been able to get Triplex.migrate to run within a transaction. Considering create_schema will do a schema drop on any {:error, some_err_message} result… not using a transaction seems an alright way of going about things.

So I would try something along the lines of:

    Triplex.create_schema(slug, Repo, fn tenant, repo ->
      {:ok, _} = Triplex.migrate(tenant, repo)
  
      Repo.transaction(fn ->
        # do other inserts etc. here... such as...
        with {:ok, user} <- insert_some_default_user_on_schema_creation(),
             {:ok, _something_else} <- do_something_else() do
          {:ok, user}
        else
          {:error, error} ->
            Repo.rollback(error)
        end
      end)
    end)

Where Next? Top

Trending in Questions Top

RSP87
I’m working on a project that simulates the bumbl example in the programming phoenix book. It acts almost like an email client. We have a...
New
kszambelanczyk
Hello! Could someone please give me a help/sample code, how to delete a file from s3 using waffle/waffle_ecto from Phoenix app. I creat...
New
RemyXRenard
I’m seeing that a list inside a Kino.DataTable will be interpreted as a charlist, even if the Kino.configure() is set to charlists: :as_l...
New
velrest
So my question is quite simple and i have found no conclusive answer on forum, google or AI. Should we use :erlang.float for Integer to ...
New
samoloth
Hi, I’ve just set up an application with ash_authentication. There is only magic link strategy for now, so there is no confirmation add o...
New
FlyingNoodle
If a change or preparation module uses Ash.Changeset.get_argument/2 or Ash.Query.get_argument/2 (or any of the other get_argument functio...
New
ryanwinchester
apply_graft/2 doesn’t rewrite an add_many sub-workflow’s deps on an add step. Grafted jobs cancel with “upstream job was deleted” Version...
New

Other Trending Topics Top

mudasobwa
I am happy to introduce the very α version of the new programming language compiled to BEAM. Welcome Cure. It has literally three kille...
New
garrison
Hobbes is a low-level distributed database for the Elixir programming language. Hobbes provides a simple, safe, and scalable storage lay...
New
marciok
Hi there! We created Gust: A task orchestrator inspired by Airflow. For those who have never heard about Aiflow, it’s a Python-based wor...
New
jimsynz
Beam Bots (or just BB for short) is a framework for building fault-tolerant robotics applications in Elixir using familiar OTP patterns. ...
New
Dmk
Xamal is a deployment tool for Elixir apps that deploys native releases to bare metal servers over SSH. It’s a port of GitHub - basecamp/...
New
Damirados
Hello everyone. After busy few months I am happy to announce v0.1.0 of Emerge &amp; Solve. They are GUI (Emerge) and State management (S...
New

We're in Beta

About us Mission Statement

Options

Thread Display Mode




Thread Preview

Skip Thread Previews