Migrate to a Chronograf HA configuration

Use chronoctl to migrate your Chronograf configuration store from BoltDB to a shared etcd data store used for Chronograf high-availability (HA) clusters.

1. Stop the Chronograf server by killing the chronograf process.

2. To prevent data loss, we strongly recommend that you back up your Chronograf data store before migrating to a Chronograf cluster.

3. Install and start etcd.

4. Run the following command, specifying the local BoltDB file and the etcd endpoint beginning with etcd://. (We recommend adding the prefix bolt:// to an absolute path. Do not use the prefix to specify a relative path to the BoltDB file.)

   chronoctl migrate \
     --from bolt:///path/to/chronograf-v1.db \
     --to etcd://localhost:2379

   Provide etcd authentication credentials

   If authentication is enabled on etcd, use the standard URI basic authentication format to define a username and password. For example:

   etcd://username:password@localhost:2379

   Provide etcd TLS credentials

   If TLS is enabled on etcd, provide your TLS certificate credentials using the following query parameters in your etcd URL:

   • cert: Path to client certificate file or PEM file
   • key: Path to client key file
   • ca: Path to trusted CA certificates

   etcd://127.0.0.1:2379?cert=/tmp/client.crt&key=/tst/client.key&ca=/tst/ca.crt

5. Update links to Chronograf (for example, from external sources) to reflect your new URLs:

   • from BoltDB: http://localhost:8888/sources/1/status
   • to etcd: http://localhost:8888/sources/373921399246786560/status

6. Set up a load balancer for Chronograf.

7. Start Chronograf.