diff --git a/docs/changelog/CHANGELOG.md b/docs/changelog/CHANGELOG.md
index 926f680a7..830de5a39 100644
--- a/docs/changelog/CHANGELOG.md
+++ b/docs/changelog/CHANGELOG.md
@@ -25,6 +25,7 @@ See also [LTS releases](https://docs.victoriametrics.com/lts-releases/).
 * FEATURE: [stream aggregation](https://docs.victoriametrics.com/stream-aggregation/): add `ignore_first_sample_interval` param to [aggregation config](https://docs.victoriametrics.com/stream-aggregation/#stream-aggregation-config). It allows users to control the time interval when aggregation skips sending aggregated samples to avoid unexpected spikes in values. By default, this interval is set to x2 of `staleness_interval`. The new setting is applicable only to `total`, `total_prometheus`, `increase`, `increase_prometheus` and `histogram_bucket` outputs. See [this issue](https://github.com/VictoriaMetrics/VictoriaMetrics/issues/7116) for details. Thanks to @iyuroch for the [pull request](https://github.com/VictoriaMetrics/VictoriaMetrics/pull/7313).
 
 * BUGFIX: [vmagent](https://docs.victoriametrics.com/vmagent): Properly return `200 OK` HTTP status code when importing data via [Pushgateway protocol](https://docs.victoriametrics.com/#how-to-import-data-in-prometheus-exposition-format) using [multitenant URL format](https://docs.victoriametrics.com/cluster-victoriametrics/#url-format). See [this issue](https://github.com/VictoriaMetrics/VictoriaMetrics/issues/3636) and [this pull request](https://github.com/VictoriaMetrics/VictoriaMetrics/pull/7571).
+* BUGFIX: [vmagent](https://docs.victoriametrics.com/vmagent): Properly set `TCP` connection timeout for `Kubernetes API server` connection for metric scrapping with `kubernetes_sd_configs`. See [this issue](https://github.com/VictoriaMetrics/VictoriaMetrics/issues/7127).
 * BUGFIX: [vmsingle](https://docs.victoriametrics.com/single-server-victoriametrics/), `vmselect` in [VictoriaMetrics cluster](https://docs.victoriametrics.com/cluster-victoriametrics/): properly return result for binary operation `^` aka pow at query requests for `NaN` values. See [this issue](https://github.com/VictoriaMetrics/VictoriaMetrics/issues/7359) for details.
 * BUGFIX: [vmui](https://docs.victoriametrics.com/#vmui): fix rendering of isolated data points on the graph that are not connected to other points.
 * BUGFIX: [vmalert](https://docs.victoriametrics.com/vmalert): improve the correctness of alert [state restoration](https://docs.victoriametrics.com/vmalert/#alerts-state-on-restarts). Previously, it could result in false-positive alerts if alert was resolved shortly before vmalert restart.
diff --git a/lib/promscrape/discovery/kubernetes/api_watcher.go b/lib/promscrape/discovery/kubernetes/api_watcher.go
index 58f16b735..e93a4d9d1 100644
--- a/lib/promscrape/discovery/kubernetes/api_watcher.go
+++ b/lib/promscrape/discovery/kubernetes/api_watcher.go
@@ -23,6 +23,7 @@ import (
 
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/logger"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/netutil"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/promauth"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/promutils"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/timerpool"
@@ -267,6 +268,7 @@ func getHTTPClient(ac *promauth.Config, proxyURL *url.URL) *http.Client {
 	getTransport := func(cfg *tls.Config) http.RoundTripper {
 		return &http.Transport{
 			Proxy:               proxy,
+			DialContext:         netutil.Dialer.DialContext,
 			TLSHandshakeTimeout: 10 * time.Second,
 			IdleConnTimeout:     *apiServerTimeout,
 			MaxIdleConnsPerHost: 100,