Snowflake receiver: conversion error for 'bytesInserted' trying to convert float64 to int64

[atoulme]

Component(s)

No response

What happened?

In snowflake receiver, for one of the snowflake pipe metric

we have the data type defined as int https://github.com/open-telemetry/opentelemetry-collector-contrib/blob/main/receiver/snowflakereceiver/model.go#L89

but in reality when parsing the SQL query https://github.com/open-telemetry/opentelemetry-collector-contrib/blob/main/receiver/snowflakereceiver/client.go#L24 and scanning the related column value sum(bytes_inserted) the receiver gets float value but expects int and hence the error.

May 19 13:05:16 MEIPLAP01089 otelcol[1304954]: 2025-05-19T13:05:16.188+0200        error        scraperhelper@v0.125.0/obs_metrics.go:61        Error scraping metrics        {"otelcol.component.id": "snowflake/azure_ci_prod_xs", "otelcol.component.kind": "receiver", "otelcol.signal": "metrics", "error": "sql: Scan error on column index 2, name \"SUM(BYTES_INSERTED)\": converting driver.Value type float64 (\"4.8225792e+08\") to a int64: invalid syntax"}
We might have to make additional checks for any other such metrics where expect int but actually get float.

Another underlying issue:

Just the presence of snowflake receiver will throw this error because it seems the receiver issues query and parses all metrics whether or not that metric is enabled. It is only after the scanning and parsing of the metrics, metric is checked whether it is enabled or not and if not enabled it is not added to the list of datapoints to be sent.

Steps to reproduce:

1. snowflake receiver

Expected:
No scan error when trying to parse snowflake metrics

Actual:

"error": "sql: Scan error on column index 2, name \"SUM(BYTES_INSERTED)\": converting driver.Value type float64 (\"4.8225792e+08\") to a int64: invalid syntax"

Collector version

0.126.0

Environment information

Environment

OS: (e.g., "Ubuntu 20.04")
Compiler(if manually compiled): (e.g., "go 14.2")

OpenTelemetry Collector configuration

Log output

Additional context

No response

[github-actions]

Pinging code owners for receiver/snowflake: @dmitryax @shalper2. See Adding Labels via Comments if you do not have permissions to add labels yourself. For example, comment '/label priority:p2 -needs-triaged' to set the priority and remove the needs-triaged label.

[ishleenk17]

@atoulme : As discussed I would take up this issue.
Please assign it to me.

[ishleenk17]

@atoulme : The issue of having the datatypes check for all the fields is still pending here, right ?
This issue is partially fixed. So, we should keep this open ?

[ishleenk17]

This sheet has detailed analysis of all the snowflake database fields and it's corresponding fields in the receiver with their datatype.

From the above analysis the issue fields are below:

RECEIVER FIELDS	RECEIVER DATATYPE	SNOWFLAKE DATATYPE	RECEIVER CALCULATIONS	SNOWFLAKE TABLE
bytesInserted	int64	FLOAT	SUM(BYTES_INSERTED)	PIPE_USAGE_HISTORY
filesInserted	int64	VARIANT	SUM(FILES_INSERTED)
storageBytes	int64( cast from float64)	NUMBER		STORAGE_USAGE
stageBytes	int64( cast from float64)	NUMBER
failsafeBytes	int64( cast from float64)	NUMBER

1. Bytes Inserted: We should change this to float64. Changing this datatype might break the current behaviour where it had int64 datatype. (Since that might be working for integer values). But for correctness we should change this.

2. filesInserted is "VARIANT" in Snowflake. Description says its no of files loaded. So, I am not sure why VARIANT would be chosen in snowflake database. Anyhow in receiver we take it as sum(files)inserted), so IMO int64 should be good here.

3. storageBytes, stageBytes, failsafeBytes are all NUMBER datataype. So these should be float64. In this issue we rectified this issue partially by taking the output of the query as float64 and then typecasting it to int64 to avoid breaking. But, IMO giving integer instead of float values is not correct. WDYT ?

cc: @shalper2 @atoulme @dmitryax