How to use a new database on Heroku - postgresql

Originally when I provisioned a database on Heroku (free one), then I my app got a new environment variable called DATABASE_URL that points to a newly created database and my app uses this one.
Today when I provisioned a new database on Heroku (paid) via the UI, I got a new environment variable called HEROKU_POSTGRESQL_MAUVE_URL that points to new database.
How do I point my Heroku app to this newly provisioned database HEROKU_POSTGRESQL_MAUVE_URL?
I expected my database URL to be swapped with HEROKU_POSTGRESQL_MAUVE_URL but that didn't happen.

This happens when your app has multiple databases:
As part of the provisioning process, a DATABASE_URL config var is added to your app’s configuration. DATABASE_URL contains the URL your app uses to access the database. If your app already has a Heroku Postgres database and you’ve provisioned another one, this config var’s name instead has the format HEROKU_POSTGRESQL_<COLOR>_URL (for example, HEROKU_POSTGRESQL_YELLOW_URL
You can promote your new database to make it the primary database:
pg:promote updates the value of the DATABASE_URL config var with the newly promoted database’s connection string. It also creates an alternate attachment for the old primary database, assigned with a new HEROKU_POSTGRESQL_<color>_URL config var. The promotion process triggers a release and restarts the app.
For example:
heroku pg:promote HEROKU_POSTGRESQL_MAUVE_URL -a your-app
You should continue to use the DATABASE_URL environment variable to connect to your database.

Related

How can I access to a PostgreSQL DB from outside of a heroku app (for a Python app)

I have a PostgreSQL DB hosted in heroku and I want to get access from other applications which aren't hosted in heroku, but I saw in the DB settings that the credentials are not permanent.
How can I get access from other application always having updated credentials?
Heroku recommends using the Heroku CLI to fetch fresh credentials every time you run your external application:
Always fetch the database URL config var from the corresponding Heroku app when your application starts. For example, you may follow 12Factor application configuration principles by using the Heroku CLI and invoke your process like so:
DATABASE_URL=$(heroku config:get DATABASE_URL -a your-app) your_process
This way, you ensure your process or application always has correct database credentials.
In this example, your_process will see an environment variable called DATABASE_URL that is set to the same value as the DATABASE_URL config far on the Heroku app called your-app.
Since you are using Python, here is one way to access that value:
import os
database_url = os.getenv("DATABASE_URL", default="some_default_for_local_development")

Heroku: importing from S3 failing

I'm trying to import a local Postgresql database to Heroku and I'm following these steps https://devcenter.heroku.com/articles/heroku-postgres-import-export#import-to-heroku-postgres.
I have successfully:
created a dump
uploaded it to an S3 Bucket
created from AWS CLI a signed link
ran the command heroku pg:backups:restore '<SIGNED URL>' DATABASE_URL (adding -a with my app name).
The process to restore a backup starts correctly but then exits with this code:
! An error occurred and the backup did not finish.
!
! Could not initialize transfer
!
! Run heroku pg:backups:info r011 for more details.
Opening the log shows:
Database: BACKUP
Finished at: 2020-01-09 18:49:30 +0000
Status: Failed
Type: Manual
Backup Size: 0.00B (0% compression)
=== Backup Logs
2020-01-09 18:49:30 +0000 Could not initialize transfer
I've tried:
re-uploading the file to the bucket,
generating a new signed link,
putting the app in maintenance mode,
I've created a user in my IAM management service with full S3 access and saved the credentials in the app environment as from https://devcenter.heroku.com/articles/s3
Not sure where to go from here but would appreciate any help. (I'm on the hobby plan therefore I can't ask Heroku's support for help)
Edit: I also tried:
deleting and recreating the S3 Bucket
installing version 1 of the AWS CLI to see if by chance the structure of a presigned link had changed
Edit 2: Since I could not find a solution I've opted to migrate the hosting entirely on AWS for the moment
Make sure that your credentials on your machine that are stored in ~/.aws/ the default value is set to the credentials you created for your heroku configs. Then also make sure the signed url is created with those credentials and configs. I had to set my default credentials to the credentials I put in my heroku configs. Then I also had to set my default region in ~/.aws/config to match the bucket location. Should work after that.
Here are some instructions if you are on mac or linux.
Sorry Windows people. I would assume it is something similar.
Create new access id and key in IAM on AWS
Set heroku configs to use those credentials heroku config:set AWS_ACCESS_KEY_ID=xxx AWS_SECRET_ACCESS_KEY=yyy
Optional (You may have to set the bucket name in heroku config too)
On your machine set your credentials you just created to the default in ~/.aws/credentials
On your machine set your default region that corresponds to your bucket in ~/.aws/config
Create signed URL aws s3 presign s3://your-bucket-address/your-object
Run restore heroku pg:backups:restore '<SIGNED URL>' DATABASE_URL
Had the exact same error and made these 2 adjustments. In the S3 console click on the file you want to use for the backup. You should see the name fo your file followed by 4 tabs. In the General information tab, do the following:
Click on Make public to make the file available for download.
Get the URL for that object where it says URL of object
(should be something like https://mybucket.s3.amazonaws.com/my.file, you can test if it works by pasting that url in a new Chrome tab and hitting that url. That should trigger the download of your file)
Once the previous check is working you can proceed to
heroku pg:backups restore 'https://mybucket.s3.amazonaws.com/my.file' DATABASE_URL
I ran into the same issue and discovered the issue was that I had my bucket's region set as us-east rather than us-east-1.

Heroku Permanent Database Credentials

I've decided to save time on the ops side of things and move to Heroku. I'm planning to have a production dyno on Heroku with a postgres database AND another dyno that reads from the same database.
However when I opened the settings of postgres, it said:
Database Credentials
Get credentials for manual connections to this database.
Please note that these credentials are not permanent.
Heroku rotates credentials periodically and updates applications where this database is attached.
What's a good way to go about this?
From Heroku Documentation,
Credentials
Do not copy and paste database credentials to a separate environment or into your application’s code. The database URL is managed by Heroku and will change under some circumstances such as:
User initiated database credential rotations using heroku pg:credentials:rotate.
Catastrophic hardware failure leading to Heroku Postgres staff recovering your database on new hardware.
Automated failover events on HA enabled plans.
It is best practice to always fetch the database URL config var from the corresponding Heroku app when your application starts. For example, you may follow 12Factor application configuration principles by using the Heroku CLI and invoke your process like so:
DATABASE_URL=$(heroku config:get DATABASE_URL -a your-app-name) your_process
This way, you ensure your process or application always has correct database credentials.
May be attaching the same database to two heroku-apps will better suit you. In this way, pg creds will be auto-managed by heroku.
I am also using this technique. I have one client-facing app and another operation-app sharing the same database instance.
You can either do this using UI or via CLI
see Share database between 2 apps in Heroku

Rename the Amazon RDS master username

Changing the password is easily done through the console. Is there any way to change the master username after creation on RDS for PostgreSQL? If so, how?
You can't change username. You can check the following links that describe how to change master password and if Amazon adds the ability to change username you will find there:
Try to find at AWS CLI for RDS:
modify-db-instance --db-instance-identifier <value> --master-user-password (string)
--master-user-password (string)
The new password for the DB instance master user. Can be any printable
ASCII character except "/", """, or "#".
Changing this parameter does not result in an outage and the change is
asynchronously applied as soon as possible. Between the time of the
request and the completion of the request, the MasterUserPassword
element exists in the PendingModifiedValues element of the operation
response. Default: Uses existing setting
Constraints: Must be 8 to 41 alphanumeric characters (MySQL, MariaDB,
and Amazon Aurora), 8 to 30 alphanumeric characters (Oracle), or 8 to
128 alphanumeric characters (SQL Server).
The Amazon RDS Command Line Interface (CLI) has been deprecated. Instead, use the AWS CLI for RDS.
Via the AWS Management Console, choose the instance you need to reset the password for, click ‘Modify’ then choose a new master password.
If
you don’t want to use the AWS Console, you can use the
rds-modify-db-instance command (as per Amazon’s documentation for RDS)
to reset it directly, given the AWS command line tools:
rds-modify-db-instance instance-name --master-user-password
examplepassword
No. As of April 2019 one cannot reset the 'master username'.
You cannot do it directly. However you can use the database migration service from AWS:
https://aws.amazon.com/dms/
Essentially you define the current database instance as your source and the new database with the correct username as your target of the migration.
This way you migrate the data from one to another database instance. As such you can change all properties including the username.
This approach has some drawbacks:
You need to configure the migration. Which takes a bit of time.
The data is migrated. This may lead unexpected behavior since not everything is eventually migrated (e.g. views etc.)
It depends how you setup everything you may experience a downtime.
Though this may not be ideal for every use-case, I did find a workaround that allows for changing the username of the master user of an AWS RDS DB.
I am using PgAdmin4 with PostgreSQL 14 at the time of writing this answer.
Login with the master user you want to change the name of
Create a new user with the following privileges and membership
Privileges and Membership
Can login - yes
Superuser - no (not possible with a managed AWS RDS DB instance, if you need complete superuser access DO NOT use a managed AWS RDS DB)
Create roles - yes
Create databases - yes
Inherit rights from the parent roles - yes
Can initiate streaming replication and backups - no (again, not possible directly without superuser permission)
Be sure to note the password used, as you will need to access this new account at least 1 time to complete the name change
Register a server with the credentials created in step 2. Disconnect from the server but do NOT remove it! Connect to the new server created
Expand Login/Group Roles and click on the master user whom you are changing the name
Click the edit icon, edit the name, and save.
Right click the server with the master username, select Properties
Update the name under the General tab if desired
Update the username under the Connection tab to whatever you changed the master username above
Save and reconnect to the server with the master user
You have successfully updated the master user's name on a managed AWS RDS DB instance, proud of you!
As #tdubs's answer states, it is possible to change the master username for a Postgres DB instance in AWS RDS. Whether it is advisable – probably not.
Here are the SQL commands you need to issue:
Create a temporary user with the CREATEROLE privilege (while being logged in with the old master user)
CREATE ROLE temp_master PASSWORD '<temporary password>' LOGIN CREATEROLE;
Now connect to the database with the temp_master user
ALTER ROLE "<old_master_username>" RENAME TO "<new_master_username>";
-- NOTICE: MD5 password cleared because of role rename
ALTER ROLE "<new_master_username>" PASSWORD '<new password>';
Now connect to the database with the <new_master_username> user in order to clean up the temporary role
DROP ROLE temp_master;
And you're done!
Warning
AWS RDS does not know that the master username has been changed, so it will keep displaying the old one and assumes that is still the master username.
This means that if you use the AWS CLI or website to update the master password, it will have no effect.
And when connecting to the database with psql you'll see:
WARNING: role "<old_master_username>" does not exist

Heroku Postgres database values resetting on deployment

I am running a simple webapp on Heroku for splitting payments between housemates, and storing all payments to a Heroku PostgreSQL database.
The app works as expected while it is running, and the database is updated properly. However, after making changes to the code locally and pushing those to heroku, the database completely resets to a past instance, and all newly added rows are missing from the database.
Is this the intended behavior or am I doing something wrong?
More details:
The app is a python app using the Flask framework
If you have a working database on Heroku and want to seed it with some values, consider either using the seed file in the db folder of your Rails app or use heroku run rails c and then use the console to seed how you'd like.