how to monitor Standalone milvusdb

[hxlsky]

there are many metrics in the output,but which is the more importand metrics need to monitor?

[yhmo]

Read the description of each metric item: https://milvus.io/docs/metrics_dashboard.md#Milvus-Metrics-Dashboard
I think the latency related metrics you can interested.

[hxlsky]

i can not find the describe the metrics such as "go_gc_duration_seconds" in the url you get?

[hxlsky]

[yhmo]

Milvus is mainly written by Golang, go_gc_duration_seconds is about the go garbage collection behavior, this is mainly for developers to observe the go gc process. Normal users no need to care this metric.

[yhmo]

We don't have a solution to monitor standalone. Standalone is not recommended for production usage.
If it is important to monitor the system state in your business, then deploy a cluster.

[hxlsky]

is there a doc about deployment milvus to k8s?

[yhmo]

Deploy a cluster by operator: https://milvus.io/docs/install_cluster-milvusoperator.md
Or helm: https://milvus.io/docs/install_cluster-helm.md

[hxlsky]

what shared storage can we use in k8s? nfs?

[yhmo]

https://milvus.io/docs/prerequisite-helm.md
https://milvus.io/docs/deploy_s3.md

[hxlsky]

is there some api in pymilvus to monitor the Standalone?
which api in pymilvus to get the data such as "http://192.168.1.134:9091/metrics"

[yhmo]

Sdk doesn't have interfaces to get metrics. All the metrics are from xxx:9091/metrics.