query, rule: make endpoint discovery dynamically reloadable

* Removed previously deprecated and hidden flags to configure endpoints ( --rule, --target, ...)
* Added new flags --endpoint.sd-config, --endpoint-sd-config-reload-interval to configure a dynamic SD file
* Moved endpoint set construction into cmd/thanos/endpointset.go for a little cleanup

The new config makes it possible to also set "strict" and "group" flags on the endpoint instead
of only their addresses, making it possible to have file based service discovery for endpoint groups too.

Signed-off-by: Michael Hoffmann <[email protected]>
Signed-off-by: Michael Hoffmann <[email protected]>

CHANGELOG.md

@@ -12,32 +12,16 @@ We use *breaking :warning:* to mark changes that are not backward compatible (re
 
 ### Fixed
 
-- [#7970](https://github.com/thanos-io/thanos/pull/7970) Sidecar: Respect min-time setting.
-
 ### Added
 
 - [#7907](https://github.com/thanos-io/thanos/pull/7907) Receive: Add `--receive.grpc-service-config` flag to configure gRPC service config for the receivers.
-- [#7961](https://github.com/thanos-io/thanos/pull/7961) Store Gateway: Add `--store.posting-group-max-keys` flag to mark posting group as lazy if it exceeds number of keys limit. Added `thanos_bucket_store_lazy_expanded_posting_groups_total` for total number of lazy posting groups and corresponding reasons.
 
 ### Changed
 
 - [#7890](https://github.com/thanos-io/thanos/pull/7890) Query,Ruler: *breaking :warning:* deprecated `--store.sd-file` and `--store.sd-interval` to be replaced with `--endpoint.sd-config` and `--endpoint-sd-config-reload-interval`; removed legacy flags to pass endpoints `--store`, `--metadata`, `--rule`, `--exemplar`.
 
 ### Removed
 
-## [v0.37.2](https://github.com/thanos-io/thanos/tree/release-0.37) - 11.12.2024
-
-### Fixed
-
-- [#7970](https://github.com/thanos-io/thanos/pull/7970) Sidecar: Respect min-time setting.
-- [#7962](https://github.com/thanos-io/thanos/pull/7962) Store: Fix potential deadlock in hedging request.
-
-### Added
-
-### Changed
-
-### Removed
-
 ## [v0.37.1](https://github.com/thanos-io/thanos/tree/release-0.37) - 04.12.2024
 
 ### Fixed

cmd/thanos/endpointset.go

@@ -198,8 +198,6 @@ func setupEndpointSet(
 	dialOpts []grpc.DialOption,
 	queryConnMetricLabels ...string,
 ) (*query.EndpointSet, error) {
-	ctx, cancel := context.WithCancel(context.Background())
-
 	configProvider, err := newEndpointConfigProvider(
 		logger,
 		configFile,
@@ -248,47 +246,48 @@ func setupEndpointSet(
 		}
 		return deduplicated
 	}
-	legacyFileSDCache := cache.New()
-	{
-		var fileSD *file.Discovery
-		if len(legacyFileSDFiles) > 0 {
-			conf := &file.SDConfig{
-				Files:           legacyFileSDFiles,
-				RefreshInterval: model.Duration(legacyFileSDInterval),
-			}
-			var err error
-			if fileSD, err = file.NewDiscovery(conf, logger, conf.NewDiscovererMetrics(reg, discovery.NewRefreshMetrics(reg))); err != nil {
-				return nil, fmt.Errorf("unable to create new legacy file sd config: %w", err)
-			}
+	var fileSD *file.Discovery
+	if len(legacyFileSDFiles) > 0 {
+		conf := &file.SDConfig{
+			Files:           legacyFileSDFiles,
+			RefreshInterval: model.Duration(legacyFileSDInterval),
 		}
+		var err error
+		if fileSD, err = file.NewDiscovery(conf, logger, conf.NewDiscovererMetrics(reg, discovery.NewRefreshMetrics(reg))); err != nil {
+			return nil, fmt.Errorf("unable to create new legacy file sd config: %w", err)
+		}
+	}
+	legacyFileSDCache := cache.New()
 
-		if fileSD != nil {
-			fileSDUpdates := make(chan []*targetgroup.Group)
+	ctx, cancel := context.WithCancel(context.Background())
 
-			g.Add(func() error {
-				fileSD.Run(ctx, fileSDUpdates)
-				return nil
+	if fileSD != nil {
+		fileSDUpdates := make(chan []*targetgroup.Group)
 
-			}, func(err error) {
-				cancel()
-			})
-			g.Add(func() error {
-				for {
-					select {
-					case update := <-fileSDUpdates:
-						// Discoverers sometimes send nil updates so need to check for it to avoid panics.
-						if update == nil {
-							continue
-						}
-						legacyFileSDCache.Update(update)
-					case <-ctx.Done():
-						return nil
+		g.Add(func() error {
+			fileSD.Run(ctx, fileSDUpdates)
+			return nil
+
+		}, func(err error) {
+			cancel()
+		})
+
+		g.Add(func() error {
+			for {
+				select {
+				case update := <-fileSDUpdates:
+					// Discoverers sometimes send nil updates so need to check for it to avoid panics.
+					if update == nil {
+						continue
 					}
+					legacyFileSDCache.Update(update)
+				case <-ctx.Done():
+					return nil
 				}
-			}, func(err error) {
-				cancel()
-			})
-		}
+			}
+		}, func(err error) {
+			cancel()
+		})
 	}
 
 	{