I added the pod name as one of the Attribute then it showed the tps values of each pod. But i need one curve which should add the tps values of each pod and show the graph

I’ve used manual instrumentation for taking the input tps value as a gauge metrics and each pod is exporting via http exporter every second. But once we’re getting the graph for TPS VS time, its showing values of only 1 pod per second. I need to know the total TPS value of n pods per second and not just one pod. I tried SUM aggregation but it’s not giving me correct value. Can someone support?

You need to add pod ID to distinguish.

I added the pod name as one of the Attribute then it showed the tps values of each pod. But i need one curve which should add the tps values of each pod and show the graph

How often are you sending the data? There will be some aggregation interval such as 60, and any aggregation you do such as SUM will perform the addition of all points falling into 60 seconds. If you need per second you need to divide by 60. You can use the query builder formula for this.

@Srikanth Chekuri Thanks for the reply. Signoz side I don't want to know how many pods are running for the gauge metric, I just need the total value of all pods. Since all pods are inserting the value with same set of attributes, I believe their fingerprint in clickhouse db will be same too? How can I have signoz to aggregate values of all pods having the same timestamp_ms and fingerprint value and show me in the graph?

I believe their fingerprint in clickhouse db will be same too?

Yes and it overrides.

How can I have signoz to aggregate values of all pods having the same timestamp_ms and fingerprint value and show me in the graph?

You can either pre-aggregate in the middle and send the aggregated data to SigNoz or send the different attributes for each pod.

The total tps as per my logs is coming 700 per pod for 1 user. If there are 3 pods i can assume the tps is 2100. But in signoz Dashboard its showing only 700 for NOOP

If you do not provide a distinguishable attribute for each pod, the time series will be the same hence 700 instead of 2100.

if I do SUM aggregate, it bumps the value to 50000 - 100000 which doesn’t make sense to me.

Aggregation happens over an interval of 60s. We don’t show data for each second since there won’t be enough pixels. The sum simply shows the aggregate. If you want per second at the intervals you can use the formula and divide by interval.