garbage collecting other resources if a given resource cannot be listed or watched. Any objects in the unsynced resource type with owner references with `blockOwnerDeletion: true` will not be known to the garbage collector. Use of `blockOwnerDeletion` has always been best-effort and racy on startup and object creation, with this fix, it continues to be best-effort for resources that cannot be synced by the garbage collector controller. ([#125796](https://github.com/kubernetes/kubernetes/pull/125796),...

NAME       CLUSTER      CDS          LDS         EDS         RDS          ECDS         ISTIOD      VERSION
proxy1     cluster1     STALE        SYNCED      SYNCED      NOT SENT     SYNCED       istiod1     1.20
proxy2     cluster2     STALE        SYNCED      STALE       SYNCED       STALE        istiod2     1.19
proxy3     cluster3     NOT SENT     ERROR       STALE       NOT SENT     NOT SENT     istiod3     1.20

NAME       CLUSTER      CDS       LDS        EDS        RDS          ECDS         ISTIOD      VERSION
proxy1     cluster1     STALE     SYNCED     SYNCED     NOT SENT     NOT SENT     istiod1     1.20

  protected Table<String, Integer, Character> create(@Nullable Object... data) {
    TestTable<String, Integer, Character> table = new TestTable<>();
    Table<String, Integer, Character> synced = Synchronized.table(table, table.mutex);
    populate(synced, data);
    return synced;
  }

  protected Table<String, Integer, Character> create(@Nullable Object... data) {
    TestTable<String, Integer, Character> table = new TestTable<>();
    Table<String, Integer, Character> synced = Synchronized.table(table, table.mutex);
    populate(synced, data);
    return synced;
  }

A queue is used to give a few properties:
* Ability to serially process updates received from a variety of different sources. This avoids need for other synchronization mechanisms like mutexes.
* Correctness at startup; with the sequencing above, items are only processed once all informers are synced. This means queries will not return stale data at startup.

* They are rarely updated after creation and initial review, and then become hard to follow, especially after important decisions are made
* They are not synced with the code to reflect the eventual solution that is committed
* Google Docs is not a "code-oriented" tool, like asciidoc can be
* Review in Google Docs is not as simple as a PR code review in GitHub

## Decision

	// Take a copy of server fields.
	var tlsConfig *tls.Config
	if srv.TLSConfig != nil {
		tlsConfig = srv.TLSConfig.Clone()
	}
	handler := srv.Handler // if srv.Handler holds non-synced state -> possible data race

	// Create new HTTP listener.
	var listener *httpListener
	listener, listenErrs := newHTTPListener(
		listenCtx,
		srv.Addrs,
		srv.TCPOptions,
	)

	var interfaceFound bool

The Health Check controller additionally controls the health status of the `WorkloadEntry`. The health is reported over the XDS client and synced with the `WorkloadEntry`.

#### Webhooks

	UpdateOpts UpdateMetadataOpts `msg:"uo"`
	FI         FileInfo           `msg:"fi"`
}

// UpdateMetadataOpts provides an optional input to indicate if xl.meta updates need to be fully synced to disk.
type UpdateMetadataOpts struct {
	NoPersistence bool `msg:"np"`
}

// CheckPartsHandlerParams are parameters for CheckPartsHandler
type CheckPartsHandlerParams struct {
	DiskID   string   `msg:"id"`