Keda Kafka lagThreshold error in parsing

[altrack]

Report

When using a Keda Kafka scaler with lagThreshold argument as 1000000 I'm getting the following error:

keda-operator  error parsing kafka metadata: error parsing lagThreshold: strconv.ParseInt: parsing "1e+06": invalid 

syntax

### Expected Behavior

Should parse numbers correctly.

### Actual Behavior

Not able to parse or parsing as: `1e+06`

### Steps to Reproduce the Problem

1. set lagThreshold as 1000000
2. kubectl  describe scaledobjects.keda.sh <your-topic>


### Logs from KEDA operator

_No response_

### KEDA Version

None

### Kubernetes Version

None

### Platform

Google Cloud

### Scaler Details

Kafka

### Anything else?

If the bug is related to another repo please let me know, I'll move it.

[dttung2905]

Hi @altrack seems like the number 1000000 is converted to scientific notation which makes strconv.ParseInt throw error. I have a few questions

• Did you wrap the deployment of scaleObject around another helm deployment ? Helm has been (in)famous for that link
• What is the KEDA version you are using? I'm testing with a latest version of KEDA from main and it works fine. I also added a test case into this and run locally but that new test case also passes 🤔

  keda/pkg/scalers/apache_kafka_scaler_test.go

  Line 64 in c5fa2ba

  var parseApacheKafkaMetadataTestDataset = []parseApacheKafkaMetadataTestData{

• May be you can try putting the number 1000000 in quotes as string instead of int type?

[zroubalik]

@dttung2905 thanks for the investigation, could you please open a PR with that test case, just to be sure? We should test this for both Kafka scalers (sarama and go based).

[dttung2905]

Hello @altrack , did you manage to resolve the issue by following the suggestion above?

[altrack]

Hey @dttung2905 , thank you for checking this and writing the test.
Unfortunately, I don't know the answers yet, it will take some time to check this.
I'll try to use a string instead.

[dttung2905]

Hey @dttung2905 , thank you for checking this and writing the test. Unfortunately, I don't know the answers yet, it will take some time to check this. I'll try to use a string instead.

Yes, please use a string instead as it might be converted to scientific notation 😅 . Do let us know if it works

[altrack]

Sorry for the delay.
Yes, changing the number to string worked as expected.
The issue was caused since in one of the Helm definition I had:
lagThreshold: {{ .Values.keda.kafka.lagThreshold | quote }},
so this caused the conversion.
Thanks for the answers.