update wiki pages

CHANGELOG.md

@@ -16,6 +16,7 @@ The following tip changes can be tested by building VictoriaMetrics components f
 ## tip
 
 * FEATURE: [vmctl](https://docs.victoriametrics.com/vmctl.html): `vmctl` `vm-native` mode now splits the migration process on per-metric basis. This allows to migrate metrics one-by-one according to the specified filter. This change allows to retry export/import requests for a specific metric and provides a better understanding of the migration progress. See [this feature request](https://github.com/VictoriaMetrics/VictoriaMetrics/issues/3600).
+* FEATURE: [vmctl](https://docs.victoriametrics.com/vmctl.html): `vmctl` `vm-native` mode now supports `--vm-native-src-headers` and `--vm-native-dst-headers` which defines headers to send with each request to the corresponding source address. Thanks, @baconmania, for the pull request. See [this pull request](https://github.com/VictoriaMetrics/VictoriaMetrics/pull/3906).
 
 * BUGFIX: prevent from possible panic during [background merge process](https://docs.victoriametrics.com/#storage). It may occur in rare case and was introduced at [v1.85.0](https://docs.victoriametrics.com/CHANGELOG.html#v1850) when implementing [this feature](https://github.com/VictoriaMetrics/VictoriaMetrics/issues/3337).
 * BUGFIX: [vmui](https://docs.victoriametrics.com/#vmui): stop showing `Please enter a valid Query and execute it` error message on the first load of vmui.

operator/high-availability.md

@@ -20,7 +20,8 @@ To run VMAgent in a highly available manner you have to configure deduplication
 
 Then increase replicas for VMAgent.
 
-create `VMSingle` with dedup flag
+create `VMSingle` with dedup flag:
+
 ```yaml
 cat <<EOF | kubectl apply -f -
 apiVersion: operator.victoriametrics.com/v1beta1
@@ -33,7 +34,9 @@ spec:
     dedup.minScrapeInterval: 60s
 EOF
 ```
-create `VMAgent` with 2 replicas
+
+create `VMAgent` with 2 replicas:
+
 ```yaml
 cat <<EOF | kubectl apply -f -
 apiVersion: operator.victoriametrics.com/v1beta1
@@ -55,10 +58,12 @@ EOF
 ```
 
 ### Sharding
+
 Sharding for `VMAgent` distributes scraping between multiple deployments of `VMAgent`.
 more info https://victoriametrics.github.io/vmagent.html#scraping-big-number-of-targets
 
 Example usage:
+
 ```yaml
 
 cat <<EOF | kubectl apply -f -
@@ -77,12 +82,27 @@ spec:
   replicaCount: 2
   remoteWrite:
     - url: "http://vmsingle-example-vmsingle-persisted.default.svc:8429/api/v1/write"
+  affinity:
+    podAntiAffinity:
+      preferredDuringSchedulingIgnoredDuringExecution:
+        - podAffinityTerm:
+            labelSelector:
+              matchLabels:
+                shard-num: '%SHARD_NUM%'
+            topologyKey: kubernetes.io/hostname
 EOF
 ```
 
 This configuration produces 5 deployments with 2 replicas at each. Each deployment has its own shard num
 and scrapes only 1/5 of all targets.
 
+You can use special placeholder `%SHARD_NUM%` in any field of `VMAgent` specification. 
+and operator will replace it with current shard num of vmagent when creating deployment or statefullset for vmagent.
+
+In the example above, the `%SHARD_NUM%` placeholder is used in the `podAntiAffinity` section, 
+which recommend to scheduler that pods with the same shard num (label `shard-num` in the pod template) 
+are not deployed on the same node. You can use another `topologyKey` for availability zone or region instead of nodes. 
+
 ### StatefulMode
 
 In `StatefulMode` `VMAgent` doesn't lose state of the PersistentQueue (file-based buffer size for unsent data) on pod restarts.
@@ -90,6 +110,7 @@ In `StatefulMode` `VMAgent` doesn't lose state of the PersistentQueue (file-base
 Operator creates `StatefulSet` and, with provided `PersistenVolumeClaimTemplate` at `StatefulStorage` configuration param, metrics queue is stored on disk.
 
 Example of configuration:
+
 ```yaml 
 apiVersion: operator.victoriametrics.com/v1beta1
 kind: VMAgent
@@ -117,6 +138,7 @@ Note, if you want to use `VMAlert` with high-available `VMAlertmanager`, which h
 at `VMAlert.spec.notifiers.[url]`. Or you can use service discovery for notifier, examples:
 
 alertmanager:
+
 ```yaml
 apiVersion: v1
 kind: Secret
@@ -155,6 +177,7 @@ spec:
   configNamespaceSelector: {}
 ```
 vmalert with fqdns:
+
 ```yaml
 apiVersion: operator.victoriametrics.com/v1beta1
 kind: VMAlert
@@ -170,6 +193,7 @@ spec:
 ```
 
 vmalert with service discovery:
+
 ```yaml
 apiVersion: operator.victoriametrics.com/v1beta1
 kind: VMAlert

vmctl.md

@@ -790,6 +790,8 @@ and specify `accountID` param.
 7. `vmctl` supports `--vm-concurrency` which controls the number of concurrent workers that process the input from source query results.
 Please note that each import request can load up to a single vCPU core on VictoriaMetrics. So try to set it according
 to allocated CPU resources of your VictoriaMetrics installation.
+8. `vmctl` supports `--vm-native-src-headers` and `--vm-native-dst-headers` which defines headers to send with each request
+to the corresponding source address.
 
 In this mode `vmctl` acts as a proxy between two VM instances, where time series filtering is done by "source" (`src`)
 and processing is done by "destination" (`dst`). So no extra memory or CPU resources required on `vmctl` side. Only