https://pinot.apache.org/ logo
Join SlackCommunities
Powered by
Hello, I'm running pinot on gke and currently duri...
# troubleshooting
t
Hello, I'm running pinot on gke and currently during query time the cpu usage never ramps up; whereas on a bare metal installation with the same setup it was 100% all the time. Do you know why this would be? I'm running 0.11.0
m
Are the results as expected?
t
yes
let me doublecheck tho
m
May be you have other processes running on bare metal that might be eating up the cpu?
Also, is this apples-apples comparison?
t
Schema + Table + data are the same
hardware is different but comparable
m
What about cpu/mem jvm other configs
t
I was talking about this with Karik,
Copy code
"mapFields": {
    "SYSTEM_RESOURCE_INFO": {
      "numCores": "1",
this was in the instance config of the servers until I've specified
Copy code
resources:
    requests:
      cpu: 7
in the helm chart
regarding mem everything is setup correctly
yes that was it
thanks
Maybe it would be nice to specify it on the chart in a comment
a
digging up this old thread; we're experiencing something potentially similar; from within the Pinot server container deployed on Kubernetes EKS, 
lscpu
returns:
Copy code
CPU(s):                          16
On-line CPU(s) list:             0-15
Thread(s) per core:              2
Core(s) per socket:              8
Socket(s):                       1
But, like Tommaso, our 
numCores
in our server config is just 
1
, which obviously limits our query performance a lot
we dont have a specific cpu resource request set on the pod, but we can try that
I guess java just outputs:
Copy code
root@pinot-server-0:/opt/pinot# java Example
Start...
Number of available processors are: 1
Open in Slack
PreviousNext
Powered by