Migrating data to GitHub Enterprise Server

Preparing the migrated data

1. Using the scp command, copy the migration archive generated from your source instance or organization to your GitHub Enterprise Server target:

   scp -P 122 PATH-TO-MIGRATION-GUID.tar.gz admin@HOSTNAME:/home/admin/
   

2. SSH into your target GitHub Enterprise Server instance. For more information, see "Accessing the administrative shell (SSH)."

   ssh -p 122 admin@HOSTNAME
   

3. Use the ghe-migrator prepare command to prepare the archive for import on the target instance and generate a new Migration GUID for you to use in subsequent steps:

   ghe-migrator prepare /home/admin/MIGRATION-GUID.tar.gz
   

   • To start a new import attempt, run ghe-migrator prepare again and get a new Migration GUID.
   • To specify where migration files should be staged append the command with --staging-path=/full/staging/path. Defaults to /data/user/tmp.

Generating a list of migration conflicts

1. Using the ghe-migrator conflicts command with the Migration GUID, generate a conflicts.csv file:

   ghe-migrator conflicts -g MIGRATION-GUID > conflicts.csv
   

   • If no conflicts are reported, you can safely import the data.

2. If there are conflicts, using the scp command, copy conflicts.csv to your local computer:

   scp -P 122 admin@HOSTNAME:conflicts.csv ~/Desktop
   

3. Continue to "Resolving migration conflicts or setting up custom mappings".

Reviewing migration conflicts

1. Using a text editor or CSV-compatible spreadsheet software, open conflicts.csv.
2. With guidance from the examples and reference tables below, review the conflicts.csv file to ensure that the proper actions will be taken upon import.

The conflicts.csv file contains a migration map of conflicts and recommended actions. A migration map lists out both what data is being migrated from the source, and how the data will be applied to the target.

model_name	source_url	target_url	recommended_action
user	https://example-gh.source/octocat	https://example-gh.target/octocat	map
organization	https://example-gh.source/octo-org	https://example-gh.target/octo-org	map
repository	https://example-gh.source/octo-org/widgets	https://example-gh.target/octo-org/widgets	rename
team	https://example-gh.source/orgs/octo-org/teams/admins	https://example-gh.target/orgs/octo-org/teams/admins	merge
project	https://example-gh.source/octo-org/widgets/projects/1	https://example-gh.target/octo-org/projects/1	merge

Each row in conflicts.csv provides the following information:

Name	Description
model_name	The type of data being changed.
source_url	The source URL of the data.
target_url	The expected target URL of the data.
recommended_action	The preferred action ghe-migrator will take when importing the data.

Possible mappings for each record type

There are several different mapping actions that ghe-migrator can take when transferring data:

action	Description	Applicable models
import	(default) Data from the source is imported to the target.	All record types
map	Instead of creating a new model based on the source data, an existing record in the target is used. Useful for importing a repository into an existing organization or mapping user identities in the target to user identities in the source.	Users, organizations, projects
rename	Data from the source is renamed, then copied over to the target.	Users, organizations, repositories, projects
map_or_rename	If the target exists, map to that target. Otherwise, rename the imported model.	Users
merge	Data from the source is combined with existing data on the target.	Teams, projects

We strongly suggest you review the conflicts.csv file and use ghe-migrator audit to ensure that the proper actions are being taken. If everything looks good, you can continue.

Resolving migration conflicts or setting up custom mappings

If you believe that ghe-migrator will perform an incorrect change, you can make corrections by changing the data in conflicts.csv. You can make changes to any of the rows in conflicts.csv.

For example, let's say you notice that the octocat user from the source is being mapped to octocat on the target.

model_name	source_url	target_url	recommended_action
user	https://example-gh.source/octocat	https://example-gh.target/octocat	map

You can choose to map the user to a different user on the target. Suppose you know that octocat should actually be monalisa on the target. You can change the target_url column in conflicts.csv to refer to monalisa.

model_name	source_url	target_url	recommended_action
user	https://example-gh.source/octocat	https://example-gh.target/monalisa	map

As another example, if you want to rename the octo-org/widgets repository to octo-org/amazing-widgets on the target instance, change the target_url to octo-org/amazing-widgets and the recommend_action to rename.

model_name	source_url	target_url	recommended_action
repository	https://example-gh.source/octo-org/widgets	https://example-gh.target/octo-org/amazing-widgets	rename

Adding custom mappings

A common scenario during a migration is for migrated users to have different usernames on the target than they have on the source.

Given a list of usernames from the source and a list of usernames on the target, you can build a CSV file with custom mappings and then apply it to ensure each user's username and content is correctly attributed to them at the end of a migration.

You can quickly generate a CSV of users being migrated in the CSV format needed to apply custom mappings by using the ghe-migrator audit command:

ghe-migrator audit -m user -g MIGRATION-GUID > users.csv

Now, you can edit that CSV and enter the new URL for each user you would like to map or rename, and then update the fourth column to have map or rename as appropriate.

For example, to rename the user octocat to monalisa on the target https://example-gh.target you would create a row with the following content:

model_name	source_url	target_url	state
user	https://example-gh.source/octocat	https://example-gh.target/monalisa	rename

The same process can be used to create mappings for each record that supports custom mappings. For more information, see our table on the possible mappings for records.

Applying modified migration data

1. After making changes, use the scp command to apply your modified conflicts.csv (or any other mapping .csv file in the correct format) to the target instance:

   scp -P 122 ~/Desktop/conflicts.csv admin@HOSTNAME:/home/admin/
   

2. Re-map the migration data using the ghe-migrator map command, passing in the path to your modified .csv file and the Migration GUID:

   ghe-migrator map -i conflicts.csv -g MIGRATION-GUID
   

3. If the ghe-migrator map -i conflicts.csv -g MIGRATION-GUID command reports that conflicts still exist, run through the migration conflict resolution process again.

Applying the imported data on GitHub Enterprise Server

1. SSH into your target GitHub Enterprise Server instance. For more information, see "Accessing the administrative shell (SSH)."

   ssh -p 122 admin@HOSTNAME
   

2. Using the ghe-migrator import command, start the import process. You'll need:

   • Your Migration GUID. For more information, see "Preparing the migrated data for import to GitHub Enterprise Server."
   • Your personal access token for authentication. The personal access token that you use is only for authentication as a site administrator, and does not require any specific scope or permissions. For more information, see "Managing your personal access tokens."

   $ ghe-migrator import /home/admin/MIGRATION-GUID.tar.gz -g MIGRATION-GUID -u USERNAME -p TOKEN
   
   > Starting GitHub::Migrator
   > Import 100% complete /
   

   • To specify where migration files should be staged append the command with --staging-path=/full/staging/path. Defaults to /data/user/tmp.

Reviewing migration data

By default, ghe-migrator audit returns every record. It also allows you to filter records by:

• The types of records.
• The state of the records.

The record types match those found in the migrated data.

Record type filters

Record type	Filter name
Users	user
Organizations	organization
Repositories	repository
Teams	team
Milestones	milestone
Projects (classic)	project
Issues	issue
Issue comments	issue_comment
Pull requests	pull_request
Pull request reviews	pull_request_review
Commit comments	commit_comment
Pull request review comments	pull_request_review_comment
Releases	release
Actions taken on pull requests or issues	issue_event
Protected branches	protected_branch

Record state filters

Record state	Description
export	The record will be exported.
import	The record will be imported.
map	The record will be mapped.
rename	The record will be renamed.
merge	The record will be merged.
exported	The record was successfully exported.
imported	The record was successfully imported.
mapped	The record was successfully mapped.
renamed	The record was successfully renamed.
merged	The record was successfully merged.
failed_export	The record failed to export.
failed_import	The record failed to be imported.
failed_map	The record failed to be mapped.
failed_rename	The record failed to be renamed.
failed_merge	The record failed to be merged.

Filtering audited records

With the ghe-migrator audit command, you can filter based on the record type using the -m flag. Similarly, you can filter on the import state using the -s flag. The command looks like this:

ghe-migrator audit -m RECORD_TYPE -s STATE -g MIGRATION-GUID

For example, to view every successfully imported organization and team, you would enter:

$ ghe-migrator audit -m organization,team -s mapped,renamed -g MIGRATION-GUID
> model_name,source_url,target_url,state
> organization,https://gh.source/octo-org/,https://ghe.target/octo-org/,renamed

We strongly recommend auditing every import that failed. To do that, you will enter:

$ ghe-migrator audit -s failed_import,failed_map,failed_rename,failed_merge -g MIGRATION-GUID
> model_name,source_url,target_url,state
> user,https://gh.source/octocat,https://gh.target/octocat,failed
> repository,https://gh.source/octo-org/octo-project,https://ghe.target/octo-org/octo-project,failed

If you have any concerns about failed imports, you can contact us by visiting GitHub Enterprise Support.

Completing the import on GitHub Enterprise Server

After your migration is applied to your target instance and you have reviewed the migration, you''ll unlock the repositories and delete them off the source. Before deleting your source data we recommend waiting around two weeks to ensure that everything is functioning as expected.

Unlocking repositories on the target instance

1. SSH into GitHub.com. If your instance comprises multiple nodes, for example if high availability or geo-replication are configured, SSH into the primary node. If you use a cluster, you can SSH into any node. Replace HOSTNAME with the hostname for your instance, or the hostname or IP address of a node. For more information, see "Accessing the administrative shell (SSH)."

   Shell

   ssh -p 122 admin@HOSTNAME
   

2. Unlock all the imported repositories with the ghe-migrator unlock command. You'll need your Migration GUID:

$ ghe-migrator unlock -g MIGRATION-GUID
> Unlocked octo-org/octo-project

Unlocking repositories on the source

After your migration is complete, you should unlock the repositories on the source.

Unlocking repositories from an organization on GitHub.com

To unlock the repositories on a GitHub.com organization, you'll send a DELETE request to the migration unlock endpoint. You'll need:

• Your access token for authentication
• The unique id of the migration
• The name of the repository to unlock

curl -H "Authorization: Bearer GITHUB_ACCESS_TOKEN" -X DELETE \
  -H "Accept: application/vnd.github.wyandotte-preview+json" \
  https://api.github.com/orgs/ORG-NAME/migrations/ID/repos/REPO_NAME/lock

Deleting repositories from an organization on GitHub.com

After unlocking the GitHub.com organization's repositories, you should delete every repository you previously migrated using the repository delete endpoint. You'll need your access token for authentication:

curl -H "Authorization: Bearer GITHUB_ACCESS_TOKEN" -X DELETE \
  https://api.github.com/repos/ORG-NAME/REPO_NAME

Unlocking repositories from a GitHub Enterprise Server instance

1. SSH into GitHub.com. If your instance comprises multiple nodes, for example if high availability or geo-replication are configured, SSH into the primary node. If you use a cluster, you can SSH into any node. Replace HOSTNAME with the hostname for your instance, or the hostname or IP address of a node. For more information, see "Accessing the administrative shell (SSH)."

   Shell

   ssh -p 122 admin@HOSTNAME
   

2. Unlock all the imported repositories with the ghe-migrator unlock command. You'll need your Migration GUID:

$ ghe-migrator unlock -g MIGRATION-GUID
> Unlocked octo-org/octo-project