Struct mz_orchestratord::controller::materialize::Context

pub struct Context {
    config: Args,
    tracing: TracingCliArgs,
    orchestratord_namespace: String,
    metrics: Arc<Metrics>,
    needs_update: Arc<Mutex<BTreeSet<String>>>,
}

Fields

config: Args

tracing: TracingCliArgs

orchestratord_namespace: String

metrics: Arc<Metrics>

needs_update: Arc<Mutex<BTreeSet<String>>>

Implementations

impl Context

pub fn new( config: Args, tracing: TracingCliArgs, orchestratord_namespace: String, metrics: Arc<Metrics>, ) -> Self

fn set_needs_update(&self, mz: &Materialize, needs_update: bool)

async fn update_status( &self, mz_api: &Api<Materialize>, mz: &Materialize, status: MaterializeStatus, needs_update: bool, ) -> Result<Materialize, Error>

Trait Implementations

impl Context for Context

type Resource = Materialize

The type of Kubernetes resource that will be watched by the Controller this context is passed to

type Error = Error

The error type which will be returned by the apply and cleanup methods

const FINALIZER_NAME: &'static str = "orchestratord.materialize.cloud/materialize"

The name to use for the finalizer. This must be unique across controllers - if multiple controllers with the same finalizer name run against the same resource, unexpected behavior can occur.

fn apply<'life0, 'life1, 'async_trait>( &'life0 self, client: Client, mz: &'life1 Self::Resource, ) -> Pin<Box<dyn Future<Output = Result<Option<Action>, Self::Error>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait,

This method is called when a watched resource is created or updated. The Client used by the controller is passed in to allow making additional API requests, as is the resource which triggered this event. If this method returns Some(action), the given action will be performed, otherwise if None is returned, success_action will be called to find the action to perform.

fn cleanup<'life0, 'life1, 'async_trait>( &'life0 self, _client: Client, mz: &'life1 Self::Resource, ) -> Pin<Box<dyn Future<Output = Result<Option<Action>, Self::Error>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait,

This method is called when a watched resource is marked for deletion. The Client used by the controller is passed in to allow making additional API requests, as is the resource which triggered this event. If this method returns Some(action), the given action will be performed, otherwise if None is returned, success_action will be called to find the action to perform.

fn success_action(&self, resource: &Self::Resource) -> Action

This method is called when a call to apply or cleanup returns Ok(None). It should return the default Action to perform. The default implementation will requeue the event at a random time between 40 and 60 minutes in the future.

fn error_action( self: Arc<Self>, resource: Arc<Self::Resource>, err: &Error<Self::Error>, consecutive_errors: u32, ) -> Action

This method is called when a call to apply or cleanup returns Err. It should return the default Action to perform. The error returned will be passed in here, as well as a count of how many consecutive errors have happened for this resource, to allow for an exponential backoff strategy. The default implementation uses exponential backoff with a max of 256 seconds and some added randomization to avoid thundering herds.