* refactor(init): Allow kubeadm init on controlplane
This shifts the cluster formation from init(bootstrap) and join(control plane)
to init(control plane).
This makes use of the previously implemented initToken to provide a TTL for
cluster initialization to take place and allows us to mostly treat all control
plane nodes equal. This also sets up the path for us to handle master upgrades
and not be concerned with odd behavior when upgrading the previously defined
init node.
To facilitate kubeadm init across all control plane nodes, we make use of the
initToken to run `kubeadm init phase certs` command to generate any missing
certificates once. All other control plane nodes will attempt to sync the
necessary certs/files via all defined trustd endpoints and being the startup
process.
* feat(init): Add service runner context to PreFunc
Signed-off-by: Brad Beam <brad.beam@talos-systems.com>
This PR introduces dependencies between the services. Now each service
has two virtual events associated with it: 'up' (running and healthy)
and 'down' (finished or failed). These events are used to establish
correct order via conditions abstraction.
Service image unpacking was moved into 'pre' stage simplifying
`init/main.go`, service images are now closer to the code which runs the
service itself.
Step 'pre' now runs after 'wait' step, and service dependencies are now
mixed into other conditions of 'wait' step on startup.
Signed-off-by: Andrey Smirnov <smirnov.andrey@gmail.com>
Conditions are now implemented as interface with two methods: `Wait` for
condition to be true (cancelable via context) and 'String' which
describes what condition is waiting for.
Generic 'WaitForAll' was implemented to wait for multiple conditions at
once.
Signed-off-by: Andrey Smirnov <smirnov.andrey@gmail.com>
Most crucial changes in `init/main.go`: on shutdown now Talos tries
to stop gracefully all the services. All the shutdown paths are unified,
including poweroff, reboot and panic handling on startup.
While I was at it, I also fixed bug with containers failing to start
when old snapshot is still around.
Service lifecycle is wrapped with `ServiceRunner` object now which
handles state transitions and captures events related to state changes.
Every change goes to the log as well.
There's no way to capture service state yet, but that is planned to be
implemented as RPC API for `init` which is exposed via `osd` to `osctl`.
Future steps:
1. Implement service dependencies for correct startup order and
shutdown order.
2. Implement service health, so that we can say "start trustd when
containerd is up and healthy".
3. Implement gRPC API for init, expose via osd (service status, restart,
poweroff, ...)
4. Impement 'String()' for conditions, so that we can see what service
is waiting on right now.
Signed-off-by: Andrey Smirnov <smirnov.andrey@gmail.com>
