アップグレードの前に

Hold your horses!  Before you do anything:

1. Read the full upgrade guide and the release notes for your destination version.
   The release notes will alert you if you need to do anything special to upgrade from one release to the next.
2. Schedule an outage or maintenance window for your deployment.
   Although we designed the upgrade process to minimize downtime for your deployment, you should do this just in case.
3. Back up your deployment using either virtualization snapshots (if you're deploying on VMware) or, if you're using AWS, by stopping your servers and creating EBS snapshots.
    Tip: It's a good idea to schedule time to back up your database while all user sessions are stopped. This will be in step 2 of both OVA procedures.
4. If you're deploying on VMware, download the binaries for the destination version and check their sha against that listed in the release notes.

Upgrade a manually deployed instance using the web UI (OVA deployment)

Use this method if you prefer web UIs instead of console or terminal environments, and you deployed your Hipchat Data Center instance on VMware, you'll perform a replacement upgrade.

1. Deploy three new Hipchat Data Center nodes using the binaries for the new version.
2. Remove the old versions of the Hipchat nodes from the load balancer, and stop their virtual machines. (Stopping the virtual machines prevents user sessions from continuing on old nodes.) 
    Tip: It's a good idea to schedule time to back up your database once all user sessions are stopped.
3. Add one of the new nodes to the load balancer.
4. From a web browser, navigate to your Hipchat cluster's load balancer. The UI redirects to a configuration screen.
5. Enter the Postgres database address for your deployment.
   The Postgres database contains all of the additional configuration information that the new node needs to set itself up. 
6. Remove the new node, and repeat steps 3 through 6 for any remaining nodes.
7. Add the three new nodes to the load balancer.
8. Take your load balancer out of maintenance mode and verify that your deployment is working as expected.

Upgrade a manually deployed instance using the CLI (OVA deployment)

If you deployed your Hipchat Data Center instance on VMware, or if you created your deployment on AWS but didn't use the quickstart CloudFormation template, you'll perform a replacement upgrade.

1. Deploy three new Hipchat Data Center nodes using the binaries for the new version.
2. Remove the old versions of the Hipchat nodes from the load balancer, and stop their virtual machines. (Stopping the virtual machines prevents user sessions from continuing on old nodes.) 
3. SSH into one of the new nodes.
4. Copy the configuration.json file you used during deployment to the new node. 

5. Apply the configuration file using the following command. Replace /path/to/configuration.json with the location of the configuration file on the host.

   
   

   hipchat datacenter configure --file /path/to/configuration.json

6. Then run the following command to restart so the new configuration takes effect. 

   hipchat datacenter restart

7. From the new node, run the hipchat datacenter selfcheck  CLI to verify the configuration. If the node passes all tests, proceed to the next step.
8. Repeat steps 3 through 7 until all old nodes are replaced with new nodes.
9. Add the three new nodes to the load balancer.
10. Take your load balancer out of maintenance mode and verify that your deployment is working as expected.

Upgrade an instance on AWS 

If you created your Hipchat Data Center deployment using the AWS quickstart CloudFormation template, all you need to do is replace the existing stack with the stack for your destination version.

To replace your CloudFormation stack:

1. Navigate to the AWS CloudFormation portal.

2. Click your existing Hipchat Data Center master stack to view its details. 

   Make sure you select the master stack, and not one of the descendant stacks. You'll know the master stack because it has the shortest name.

   

3. On the details screen, click Update Stack.
   
4. On the next screen, select Specify an Amazon S3 template URL, and enter the URL for the destination version of the CloudFormation template.
   For upgrades on a production deployment, this will usually be: https://s3.amazonaws.com/hipchat-server-stable/dc/production/templates/quickstart-hipchat-master.template
   (Beta deployments may use a slightly different URL)
   
5. Click Next, and review the values the new stack provides. No changes are necessary and this screen is for confirmation only.
6. Check the box next to I acknowledge that AWS CloudFormation might create IAM resources with custom names.
7. Click Update. 
   Amazon begins the stack update. This process takes about an hour.

How upgrading an AWS CloudFormation stack works

When you replace the old stack, AWS performs the following actions in a rolling upgrade fashion:

1. Creates a new Auto Scaling Group (ASG) that specifies the new AMI.
2. Stops one of the old nodes and removes it from the load balancer.
3. Creates and configures a new node.
4. Runs an AWS healthcheck on the new node, and if it passes, adds the new node to the load balancer.
5. Repeats steps 2 through 4 until all old nodes have been replaced with new nodes. 

Verify the upgrade

Hipchat Data Center's web UI includes information about its version in the footer at the bottom of the page. Load the page for each node and confirm that the version at the bottom of the page is the one you expect. If you are using the command line, you can verify the upgrade by running the hipchat --version CLI on each node, or checking the /etc/hipchat-release on all three hosts.

If you're using AWS, log in to the Hipchat Data Center admin UI, and click System. On the screen that appears, make sure that the IPs listed for the nodes in the cluster match the IPs of the newly deployed Hipchat nodes.