Usage Question - custom_query vs get_metric_range_data

Question

I have the following custom query:

rate(memcached_evictions{environment=~'production',location=~'location1',service='service1'}[1m])

When I get the data return over raw HTTP (w/ requests.get()) I get my results back as a vector resultType with other related metric data. The values returned are what I expected:

“value”: [<float>, “0”]

Using custom_query I can obtain the same results. But when I attempt to use one of the other methods you kind folks provide, such as get_metric_range_data, the results returned are quite different than what I’d expect. What am I missing?

Here’s the code I’m using for get_metric_data_range:

start_time = parse_datetime("15m")
end_time = parse_datetime("now")

curr_range_data = prom.get_metric_range_data(
    metric_name="memcached_evictions{environment=~'production',location=~'location1',service='service1'}",
    start_time=start_time,
    end_time=end_time
)

calc_range_data = MetricSnapshotDataFrame(curr_range_data)

The data returned when printing out calc_range_data:

__name__                value
memcached_evictions ... 3452222
memcached_evictions ... 3333263
memcached_evictions ... 13454715
memcached_evictions ... 9638913

Based on the results I get back from custom_query and the Prometheus Expression Browser, the values returned here are quite different from the expected values.

Now, I feel like I’m doing something wrong here in how I’m using these pieces, I don’t believe this to be a bug in the code. Just trying to understand how to best make use of ya’lls good work 😃

Any guidance here is greatly appreciated.

[Shackelford-Arden]

@chauhankaranraj You’re correct! That does do what I need. Thank you!

I didn’t realize that internally, it was just passing in the query I provided. I thought there was some other stuff going on in the query prep piece before it was sent off to Prometheus. I’ve looked at the source code now, makes waaaaay more sense.

Thanks for the quick response @4n4nd , I don’t think I’ve ever seen such a quick response 😃

[chauhankaranraj]

Using custom_query I can obtain the same results. But when I attempt to use one of the other methods you kind folks provide, such as get_metric_range_data, the results returned are quite different than what I’d expect. What am I missing?

Hi @Shackelford-Arden, could this be because the input being passed to custom_query is different than that being passed to get_metric_range_data? I.e. the former looks something like rate ( your_query [1m] ) vs the latter is just your_query.

Does using get_current_metric_value instead like shown below give the expected results?

curr_data = prom.get_current_metric_value(
    metric_name="rate(memcached_evictions{environment=~'production',location=~'location1',service='service1'}[1m])",
    params={'time': start_time.timestamp()},
)
calc_data = MetricSnapshotDataFrame(curr_data)