Bitbucket Data Center Installation

[Terri L O'Mahony]

We are considering purchasing the data center for mirroring.  If we proceed down this path do we need to also install the bitbucket server first?

Thanks,

Terri

[Dave Theodore _Coyote Creek Consulting_ · Accepted solution]

The Bitbucket Installation Guide should cover all the major topics of setting up the application. In short, you set up one node first, and then add additional nodes.  When we work with a client  that is already using Stash or Bitbucket, we typically build a new Data Center environment and migrate their data in to the new environment during a change control.  This tends to limit the length of the downtime required to cut over to Data Center.

If your reason for using Data Center is high availability, you need to consider several things that are not really covered in the guide, though. You'll want to make sure your database is highly available. Depending on the technology you choose and your staff's proficiency with database clustering, you may want to reach out to an Atlassian Solution Partner for some help. We typically implement clustered Postgres for most clients, but if your team has more familiarity with MS SQL Server or Oracle, they can also be clustered.  Also, MySQL is not supported on Bitbucket Data Center (only,) in case you were considering using it. You will also want a clustered load balancer, and you will need one that can support normal TCP proxying, as well as HTTP(S) proxying, if you intend to have people use command line Git clients and SSH. We typically implement HAproxy and have a high availability design that we have developed.  Elasticsearch is a component of modern versions of Bitbucket and it needs to be clustered in order to be highly available. Finally, your shared storage volume also needs to come from some sort of clustered storage system.  

If you're only after scalability, then you don't need to pay as much attention to this, though.  

[Terri L O'Mahony]

Thank you so much for response.  We are currently using Bitbucket, our main reason for wanting to implement the data center (mirroring) is to gain availability.   Currently we understand that we can only obtain a backup of our code if it is a true disaster.  Meaning if, for some reason a developer would accidentally delete the code we could  not obtain a copy. 

Also, as you can see I am new to Bitbucket.  That being said, I am trying to determine what the best path would be for my organization. 

Thanks,

Terri