The DC/OS Telegraf service requires the `/var/lib/dcos/cluster/id` file to function properly, otherwise the service will fail to start and it will continuously restart.
Usually one does not see any issues if they have dcos-metrics enabled. Since dcos-metrics service is not mandatory we need a fix so dcos-telegraf service can be able to start.
I have an environment where I disable dcos-metrics (which has a bootstrap method to create the cluster-id file) and I get the following error when starting dcos-telegraf:
I propose the following:
- Add a check `ExecStartPre=/usr/bin/test -f /var/lib/dcos/cluster-id` into the systemd service file
- Implement the bootstrap methods for the telegraf service that will create the cluster-id file.