added doc for placeholder support in vmagent specification for operator (https://github.com/VictoriaMetrics/operator/issues/592) (#3916)

This commit is contained in:
Alexander Marshalov 2023-03-06 11:58:31 +01:00 committed by Aliaksandr Valialkin
parent 0efdf47571
commit c682609fc4
No known key found for this signature in database
GPG key ID: A72BEC6CD3D0DED1

View file

@ -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