Use one of the following methods (depending on the changes) to upgrade a cluster to a newer version.
Use the following procedure to rotate all cluster nodes, one server at a time:
- Add a new server to the cluster with a configuration that joins them to the existing cluster.
- Stop dkron service on one of the old servers, if it was the leader allow a new leader to be elected. Note that it is better to remove the current leader at the end, to ensure a leader is elected from the new nodes.
dkron raft list-peersto list current cluster nodes.
dkron raft remove-peerto forcefully remove the old server.
- Repeat steps above until all old cluster nodes have been upgraded.
Backup & Restore
/restore API endpoint to restore a previously exported jobs file
curl localhost:8080/v1/jobs > backup.json
curl localhost:8080/v1/restore --form '[email protected]'
This will restore all jobs and counters as they were in the export file.