query can't get result of arithmetic binary operators between two scalars

[wangyanzhao]

Thanos used: v0.10.1

Object Storage Provider: no use of object store

What happened: can't get result of arithmetic binary operators between two scalars, "No datapoints found". But it can get result from Prometheus.

What you expected to happen:

How to reproduce it (as minimally and precisely as possible):

Full logs to relevant components:

Anything else we need to know:

[GiedriusS]

What's :node_cpu_utilisation:avg1m? Does it have any label? FWIW I can query sum(up) * 2 on 0.10.1 just fine. Could you please write some instructions on how to reproduce it? Do you have Thanos Sidecar connected to your Thanos Query?

[wangyanzhao]

All separate query is ok. Such as

1. :node_cpu_utilisation:avg1m

2. sum(kube_node_status_capacity_cpu_cores)

3. :node_cpu_utilisation:avg1m * 2

[wangyanzhao]

@GiedriusS

[wangyanzhao]

query "sum(:node_cpu_utilisation:avg1m) * sum(kube_node_status_capacity_cpu_cores)" is ok.

but in Promtheus, no need sum, ":node_cpu_utilisation:avg1m * sum(kube_node_status_capacity_cpu_cores)" is ok.

[GiedriusS]

Still haven't gotten around to this exact issue but brainstormed an idea that we could run these correctness tests of PromQL in our CI: https://github.com/prometheus/prometheus/tree/master/promql/testdata on a simple Sidecar+Query setup.

[stale]

This issue/PR has been automatically marked as stale because it has not had recent activity. Please comment on status otherwise the issue will be closed in a week. Thank you for your contributions.

[bwplotka]

Did we get onto this issue eventually?

…

On Sat, 18 Apr 2020 at 15:51, stale[bot] ***@***.***> wrote: This issue/PR has been automatically marked as stale because it has not had recent activity. Please comment on status otherwise the issue will be closed in a week. Thank you for your contributions. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#2245 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABVA3O3PWPHQHLLBWMCPPDLRNG475ANCNFSM4LE55M7A> .

[GiedriusS]

I couldn't reproduce this last time I have tried.

[bwplotka]

Awesome, let's close - if anyone can repro, just comment we will reopen 🤗

[yeya24]

Still haven't gotten around to this exact issue but brainstormed an idea that we could run these correctness tests of PromQL in our CI: https://github.com/prometheus/prometheus/tree/master/promql/testdata on a simple Sidecar+Query setup.

That sounds interesting. Cortex has cortexproject/cortex#2203 to test the performance between its chunk and block storage. We can use a similar idea to do some AB testing between Prometheus and Thanos

[GiedriusS]

Still haven't gotten around to this exact issue but brainstormed an idea that we could run these correctness tests of PromQL in our CI: https://github.com/prometheus/prometheus/tree/master/promql/testdata on a simple Sidecar+Query setup.

That sounds interesting. Cortex has cortexproject/cortex#2203 to test the performance between its chunk and block storage. We can use a similar idea to do some AB testing between Prometheus and Thanos

Maybe let's put this into a separate issue? What do you think? :P

[yeya24]

Sounds good to me. Would you like to open it?
I think https://github.com/prometheus/prometheus/tree/master/promql/testdata is not sufficient to cover all cases. If there is a promql fuzzer to randomly generate query, then it would be amazing.