Migrate to could jira server with external Crowd directory

[elvis]

We are trying to help a customer to migrate Jira software and service desk to cloud. They use Crowd as the user management and SSO tool. As far as I understand from documentation, the only path is to migrate the external users to internal Jira directory first (which is tricky AND would require all users to change their passwords plus would create an issue for the connected tools such as Confluence, BitBucket etc.), then migrate to cloud and then to implement Access. 

This seems very unusual that there is no upgrade path from a standard Atlassian tool setup (Jira + Crowd + Confluence + Bitbucket) to cloud without uprooting the whole user system. Am I missing something here?

Followup question - if we would do that now (migrate Crowd users to Jira and then use Jira directory which is also crowd for Confluence/Bitbucket user management) and migrate Jira to cloud - what then? Does that mean that we would need to migrate users same way also in Confluence and Bitbucket? How does Access come into this picture?

Very confused and puzzled here so any help would be appreciated.

[Artem Fomin - DevOps Action]

Technically you should not have to migrate users to an internal directory for Cloud migration. User data is synchronized to App DB. Still, upon migration, users would need to be using their Atlassian passwords (or Google ones or whatever) which would not necessarily match their Crowd credentials.

Jira/Confluence use the same Cloud site. Bitbucket is somewhat of a separate offering, but the process is not much different user-wise.

If you really want to keep using Crowd (something I would not recommend, unless there is a business reason I am unaware of), you may be able to connect Atlassian Access to Crowd via SAML, but this feels redundant to me.

[elvis]

Hi Artem,

That’s what I expected, but unfortunately not so. If we create users separately in the Jira Internal Directory then it works OK.

elvis

[Artem Fomin - DevOps Action]

Here is what you can do:

- Pull user data from DB:

SELECT email_address AS "email", user_name AS "username", display_name AS "displayname", active AS "active" FROM cwd_user

Save it as users.csv

SELECT DISTINCT parent_name AS "groupname", child_name AS "username" FROM cwd_membership

Save it as groups.csv

- Strip user data from entities.xml:

sed -i '/<User id="/d' entities.xml
sed -i '/<Group id="/d' entities.xml
sed -i '/<Membership id="/d' entities.xml

Provide user dumps and trimmed export to Atlassian Support, and they'll import this data for you.