Upgrade on Bare Metal

This guide will help you upgrade an OpenMetadata deployment using release binaries.

Requirements

This guide assumes that you have an OpenMetadata deployment that you installed and configured following the Bare Metal deployment guide.

Prerequisites

Everytime that you plan on upgrading OpenMetadata to a newer version, make sure to go over all these steps:

Backup your Metadata

Before upgrading your OpenMetadata version we strongly recommend backing up the metadata.

The source of truth is stored in the underlying database (MySQL and Postgres supported). During each version upgrade there is a database migration process that needs to run. It will directly attack your database and update the shape of the data to the newest OpenMetadata release.

It is important that we backup the data because if we face any unexpected issues during the upgrade process, you will be able to get back to the previous version without any loss.

Since version 1.4.0, OpenMetadata encourages using the builtin-tools for creating logical backups of the metadata:

• mysqldump for MySQL
• pg_dump for Postgres

For PROD deployment we recommend users to rely on cloud services for their databases, be it AWS RDS, Azure SQL or GCP Cloud SQL.

If you're a user of these services, you can leverage their backup capabilities directly:

• Creating a DB snapshot in AWS
• Backup and restore in Azure MySQL
• About GCP Cloud SQL backup

You can refer to the following guide to get more details about the backup and restore:

Understanding the "Running" State in OpenMetadata

In OpenMetadata, the "Running" state indicates that the OpenMetadata server has received a response from Airflow confirming that a workflow is in progress. However, if Airflow unexpectedly stops or crashes before it can send a failure status update through the Failure Callback, OpenMetadata remains unaware of the workflow’s actual state. As a result, the workflow may appear to be stuck in "Running" even though it is no longer executing.

This situation can also occur during an OpenMetadata upgrade. If an ingestion pipeline was running at the time of the upgrade and the process caused Airflow to shut down, OpenMetadata would not receive any further updates from Airflow. Consequently, the pipeline status remains "Running" indefinitely.

Running State in OpenMetadata

Expected Steps to Resolve

To resolve this issue:

• Ensure that Airflow is restarted properly after an unexpected shutdown.
• Manually update the pipeline status if necessary.
• Check Airflow logs to verify if the DAG execution was interrupted.

Update sort_buffer_size (MySQL) or work_mem (Postgres)

Before running the migrations, it is important to update these parameters to ensure there are no runtime errors. A safe value would be setting them to 20MB.

If using MySQL

You can update it via SQL (note that it will reset after the server restarts):

To make the configuration persistent, you'd need to navigate to your MySQL Server install directory and update the my.ini or my.cnf files with sort_buffer_size = 20971520.

If using RDS, you will need to update your instance's Parameter Group to include the above change.

If using Postgres

You can update it via SQL (not that it will reset after the server restarts):

To make the configuration persistent, you'll need to update the postgresql.conf file with work_mem = 20MB.

If using RDS, you will need to update your instance's Parameter Group to include the above change.

Note that this value would depend on the size of your query_entity table. If you still see Out of Sort Memory Errors during the migration after bumping this value, you can increase them further.

After the migration is finished, you can revert this changes.

Backward Incompatible Changes

1.6.5

Airflow 2.10.5

We are upgrading the Ingestion Airflow version to 2.10.5.

The upgrade from the existing 2.9.1 (or 2.9.3) -> 2.10.5 should happen transparently. The only thing to note is that there's an ongoing issue with Airflow migrations and the pymysql driver, which we used before. If you are specifying on your end the DB_SCHEME environment variable in the ingestion image, make sure it now is set to mysql+mysqldb.

We have updated the default values accordingly.

1.6.4

Airflow 2.9.3

We are upgrading the Ingestion Airflow version to 2.9.3.

The upgrade from the existing 2.9.1 -> 2.9.3 should happen transparently. The only thing to note is that there's an ongoing issue with Airflow migrations and the pymysql driver, which we used before. If you are specifying on your end the DB_SCHEME environment variable in the ingestion image, make sure it now is set to mysql+mysqldb.

We have updated the default values accordingly.

1.6.2

Executable Logical Test Suites

We are introducing a new feature that allows users to execute logical test suites. This feature will allow users to run groups of Data Quality tests, even if they belong to different tables (or even services!). Note that before, you could only schedule and execute the tests for each of the tables.

From the UI, you can now create a new Test Suite, add any tests you want and create and schedule the run.

This change, however, requires some adjustments if you are directly interacting with the OpenMetadata API or if you are running the ingestions externally:

/executable endpoints Changes

CRUD operations around "executable" Test Suites - the ones directly related to a single table - were managed by the /executable endpoints, e.g., POST /v1/dataQuality/testSuites/executable. We'll keep this endpoints until the next release, but users should update their operations to use the new /base endpoints, e.g., POST /v1/dataQuality/testSuites/base.

This is to adjust the naming convention since all Test Suites are executable, so we're differentiating between "base" and "logical" Test Suites.

In the meantime, you can use the /executable endpoints to create and manage the Test Suites, but you'll get deprecation headers in the response. We recommend migrating to the new endpoints as soon as possible to avoid any issues when the /executable endpoints get completely removed.

YAML Changes

If you're running the DQ Workflows externally AND YOU ARE NOT STORING THE SERVICE INFORMATION IN OPENMETADATA, this is how they'll change:

A YAML file for 1.5.x would look like this:

Basically, if you are not storing the service connection in OpenMetadata, you could leverage the source.serviceConnection entry to pass that information.

However, with the ability to execute Logical Test Suites, you can now have multiple tests from different services! This means, that the connection information needs to be placed differently. The new YAML file would look like this:

As you can see, you can pass multiple serviceConnections to the sourceConfig entry, each one with the connection information and the serviceName they are linked to.

1.6.0

Ingestion Workflow Status

We are updating how we compute the success percentage. Previously, we took into account for partial success the results of the Source (e.g., the tables we were able to properly retrieve from Snowflake, Redshift, etc.). This means that we had an error threshold in there were if up to 90% of the tables were successfully ingested, we would still consider the workflow as successful. However, any errors when sending the information to OpenMetadata would be considered as a failure.

Now, we're changing this behavior to consider the success rate of all the steps involved in the workflow. The UI will then show more Partial Success statuses rather than Failed, properly reflecting the real state of the workflow.

Database Metadata & Lineage Workflow

With 1.6 Release we are moving the View Lineage & Stored Procedure Lineage computation from metadata workflow to lineage workflow.

This means that we are removing the overrideViewLineage property from the DatabaseServiceMetadataPipeline schema which will be moved to the DatabaseServiceQueryLineagePipeline schema.

Profiler & Auto Classification Workflow

We are creating a new Auto Classification workflow that will take care of managing the sample data and PII classification, which was previously done by the Profiler workflow. This change will allow us to have a more modular and scalable system.

The Profiler workflow will now only focus on the profiling part of the data, while the Auto Classification will take care of the rest.

This means that we are removing these properties from the DatabaseServiceProfilerPipeline schema:

• generateSampleData
• processPiiSensitive
• confidence which will be moved to the new DatabaseServiceAutoClassificationPipeline schema.

What you will need to do:

• If you are using the EXTERNAL ingestion for the profiler (YAML configuration), you will need to update your configuration, removing these properties as well.
• If you still want to use the Auto PII Classification and sampling features, you can create the new workflow from the UI.

RBAC Policy Updates for EditTags

We have given more granularity to the EditTags policy. Previously, it was a single policy that allowed the user to manage any kind of tagging to the assets, including adding tags, glossary terms, and Tiers.

Now, we have split this policy to give further control on which kind of tagging the user can manage. The EditTags policy has been split into:

• EditTags: to add tags.
• EditGlossaryTerms: to add Glossary Terms.
• EditTier: to add Tier tags.

Collate - Metadata Actions for ML Tagging - Deprecation Notice

Since we are introducing the Auto Classification workflow, we are going to remove in 1.7 the ML Tagging action from the Metadata Actions. That feature will be covered already by the Auto Classification workflow, which even brings more flexibility allow the on-the-fly usage of the sample data for classification purposes without having to store it in the database.

Service Spec for the Ingestion Framework

This impacts users who maintain their own connectors for the ingestion framework that are NOT part of the OpenMetadata python library (openmetadata-ingestion). Introducing the "connector specifcication class (ServiceSpec)". The ServiceSpec class serves as the entrypoint for the connector and holds the references for the classes that will be used to ingest and process the metadata from the source. You can see postgres for an implementation example.

Fivetran

The filtering of Fivetran pipelines now supports using their names instead of IDs. This change may affect existing configurations that rely on pipeline IDs for filtering.

DBT Cloud Pipeline Service

We are removing the field jobId which we required to ingest dbt metadata from a specific job, instead of this we added a new field called jobIds which will accept multiple job ids to ingest metadata from multiple jobs.

MicroStrategy

The serviceType for MicroStrategy connector is renamed from Mstr to MicroStrategy.

Upgrade process

Step 1: Download the binaries for the release you want to install

OpenMetadata release binaries are maintained as GitHub releases.

To download a specific release binary:

• Visit github.com/open-metadata/OpenMetadata/releases. The latest release will be at the top of this page.
• Locate the Assets' section for the release you want to upgrade to.
• Download the release binaries. The release binaries will be in a compressed tar file named using the following convention, openmetadata-x.y.z.tar.gz Where x, y, z are the major, minor, and patch release numbers, respectively.

Step 2: Extract the release binaries from the download file

Using the command-line tool or application of your choice, extract the release binaries.

For example, to extract using tar, run the following command.

This will create a directory with the same name as the download file minus the .tar and .gz extensions.

Step 3: Navigate into the directory created by extracting the release binaries

Change into the new directory by issuing a command similar to the following.

For example, to navigate into the directory created by issuing the tar command above, you would run the following command.

Step 4: Stop the OpenMetadata server

OpenMetadata ships with a few control scripts. One is openmetadata.sh. This script enables you to start, stop, and perform other deployment operations on the OpenMetadata server.

Most OpenMetadata releases will require you to migrate your data to updated schemas.

Before you migrate your data to the new release you are upgrading to, stop the OpenMetadata server from the directory of your current installation by running the following command:

Step 5: Migrate the database schemas and ElasticSearch indexes

The bootstrap/openmetadata-ops.sh script enables you to perform a number of operations on the OpenMetadata database (in MySQL) and index (in Elasticsearch).

Step 6: Restart the OpenMetadata server

Once you've dropped and recreated your data in the new version, restart the OpenMetadata server using the new release binaries. You may restart the server by running the following command.

Post-Upgrade Steps

Reindex

With UI

Go to Settings -> Applications -> Search Indexing

Reindex

Before initiating the process by clicking Run Now, ensure that the Recreate Indexes option is enabled to allow rebuilding the indexes as needed.

In the configuration section, you can select the entities you want to reindex.

Reindex

Since this is required after the upgrade, we want to reindex All the entities.

(Optional) Update your OpenMetadata Ingestion Client

If you are running the ingestion workflows externally or using a custom Airflow installation, you need to make sure that the Python Client you use is aligned with the OpenMetadata server version.

For example, if you are upgrading the server to the version x.y.z, you will need to update your client with

The plugin parameter is a list of the sources that we want to ingest. An example would look like this openmetadata-ingestion[mysql,snowflake,s3]==1.2.0. You will find specific instructions for each connector here.

Moreover, if working with your own Airflow deployment - not the openmetadata-ingestion image - you will need to upgrade as well the openmetadata-managed-apis version:

With Kubernetes

Follow these steps to reindex using the CLI:

1. List the CronJobs Use the following command to check the available CronJobs:

Upon running this command you should see output similar to the following.

2. Create a Job from a CronJob Create a one-time job from an existing CronJob using the following command:

Replace <job_name> with the actual name of the job.

Upon running this command you should see output similar to the following.

3. Check the Job Status Verify the status of the created job with:

Upon running this command you should see output similar to the following.

4. view logs To view the logs use the below command.

Replace <job_name> with the actual job name.

Re Deploy Ingestion Pipelines

With UI

Go to Settings -> {Services} -> {Databases} -> Pipelines

Re-deploy

Select the pipelines you want to Re Deploy click Re Deploy.

With Kubernetes

Follow these steps to deploy pipelines using the CLI:

1. List the CronJobs Use the following command to check the available CronJobs:

Upon running this command you should see output similar to the following.

2. Create a Job from a CronJob Create a one-time job from an existing CronJob using the following command:

Upon running this command you should see output similar to the following.

3. Check the Job Status Verify the status of the created job with:

Upon running this command you should see output similar to the following.

4. view logs To view the logs use the below command.

Replace <job_name> with the actual job name.

If you are seeing broken dags select all the pipelines from all the services and re deploy the pipelines.

Openmetadata-ops Script

Overview

The openmetadata-ops script is designed to manage and migrate databases and search indexes, reindex existing data into Elastic Search or OpenSearch, and redeploy service pipelines.

Usage

Commands

• analyze-tables

Migrates secrets from the database to the configured Secrets Manager. Note that this command does not support migrating between external Secrets Managers.

• changelog

Prints the change log of database migration.

• check-connection

Checks if a connection can be successfully obtained for the target database.

• deploy-pipelines

Deploys all the service pipelines.

• drop-create

Deletes any tables in the configured database and creates new tables based on the current version of OpenMetadata. This command also re-creates the search indexes.

• info

Shows the list of migrations applied and the pending migrations waiting to be applied on the target database.

• migrate

Migrates the OpenMetadata database schema and search index mappings.

• migrate-secrets

Migrates secrets from the database to the configured Secrets Manager. Note that this command does not support migrating between external Secrets Managers.

• reindex

Reindexes data into the search engine from the command line.

• repair

Repairs the DATABASE_CHANGE_LOG table, which is used to track all the migrations on the target database. This involves removing entries for the failed migrations and updating the checksum of migrations already applied on the target database.

• validate

Checks if all the migrations have been applied on the target database.

Examples

Display Help To display the help message:

Migrate Database Schema

To migrate the database schema and search index mappings:

Reindex Data

To reindex data into the search engine: