https://signoz.io logo
Join Slack
Powered by
hey <@U03QS81J47L> we primarily use PromQL as the ...
# support
p
hey @sudhanshu dev we primarily use PromQL as the query language You can also create dashboard with Query builder, which will create native clickhouse queries. We provided PromQL support because many users are famiiar with PromQL and sometimes have dashboards written in that
s
I explored it.Any documents
I imported basic dashboard for cpu and memory metrics provided by signoz
I am receiving only memory and cpu utilisation
No cpu limits/requests mertics
No memory limits/requests mertics
p
I see. Are you just relying on hostmetrics receiver which is enabled by default or you have also configured prometheus node exporter and scraping it via otel collector? This doc might be helpful - https://signoz.io/docs/userguide/send-metrics/
s
No Prometheus exporter
Thank you
Let me start working on it
a
cc @Prashant Shahi
s
So flow would be like this Prometheus node exporter push metrics to Prometheus and then from Prometheus we will push into signoz and then create dashboards
Is it my correct understanding?
As it is the part of kubernetes mertics
Did I missed anything?
p
right, naming convention would vary a little. Metrics exporter => Prometheus receiver (OtelCollectorMetrics) => SigNoz (powered by ClickHouse) After that, we can plot graphs in dashboard.
@sudhanshu dev did you follow the k8s infra metrics guide below? https://signoz.io/docs/tutorial/kubernetes-infra-metrics/
s
Yes
I followed the same
I imported the dashboard json provided inthe document
Cpu and memory metrics are coming
p
can you share details about your k8s cluster? like which cloud/tool to setup k8s cluster, number of nodes and node type.
s
But cpu request/limits metrics are not coming
Eks
6 nodes
a
I think @sudhanshu dev is following a different way to get node metrics, via node-exporter and not in built 
hostmetricsreceiver
@Prashant Shahi possible to get on a call sometime with @sudhanshu dev and sort out the way k8s metrics are being collected?
s
Sure we can do that
Plz let me know any possible time
Tomorrow
We can connect
And discuss
p
let's catch up tmrw at 3pm if that works out for you.
s
I am not using node exporter @Ankit Nayan
Sure
Will work for mw
Do we need to book calendar slot?
For tomorrow
p
@sudhanshu dev can you share your email id over DM?
s
Done
v
@sudhanshu dev even if they use EBS, can we see if we can use efs? There wont be any need to commit ebs volume
s
As they used pvc in there k8s deployment so there should not be any issues
In using efs
@Ankit Nayan is it correct?
a
I didn't get the context? Is this related to using hostmetricsreceiver or something else?
cc: @Prashant Shahi
s
it is related to ebs vs efs as clickhouse backend
Can we use efs instead of ebs in clickhouse
?
a
I never tested that... @Prashant Shahi have you tested this?
p
@sudhanshu dev we haven't tested EFS from our side, but it should work as it is supported by ClickHouse.
s
Same here it should work as it is independent of click house
Working it is just storage backend
p
Let's get into it over the call later. Meanwhile, you can have a look at the clickhouse benchmark for EFS and EBS. https://clickhouse.com/benchmark/hardware/#[0,[%22t3.xlarge,%20EBS%22,%22t3.xlarge,%20EFS%22],[%220%22,%221%22,%222%22]]
s
Yes
p
can you attach this in inframetrics config?
Copy code
metric_groups:
      - node
      - pod
      - container
      - volume
Open in Slack
PreviousNext
Powered by