feat: added min_over_time TraceQL metric - #3975
Conversation
mdisibio
left a comment
mdisibio
left a comment
There was a problem hiding this comment.
As discussed offline - we need to add a new aggregator for mode Sum/Final, which computes the min across time-series (i.e. the min of all mins), and use it in MetricsAggregate.initSum and initFinal. Right now it is still falling back to SimpleAdditionAggregator.
|
Back to draft, we need to come out with a solution for missing/uninitialized time-series values |
|
Is this something we'll want to document? |
Yes, once we agreed on a solution we should document this new function |
This may be a good spot for your content. The compare function content went on the traceql/metrics-queries.md file in the Functions section. |
mdisibio
left a comment
mdisibio
left a comment
There was a problem hiding this comment.
This is looking great. I'm really happy with this PR because we are straightening up some ast and value handling quirks that needed it. There is 1 more required change to drop series with no values, and then 1 small nit that I would prefer but not blocking.
@knylander-grafana I've added some documentation. Maybe is a good idea to have a section for all the aggregation functions over_time. Like prometheus does |
|
If you have a PR with doc updates, add the type/docs label to it please. |
@javiermolinar Thank you for adding the doc! We have a doc issue to update the metrics query docs (they need some love). Before we restructure sections in the page, we should look at the content on the metrics query pages to consider who our target audience is, what are the topics that audience needs to know, what are they interested in doing (determine any tasks -- which helps us create meaningful examples), etc. The Prometheus example is a good way of just listing an item and defining what it does. It could be a good way to do an introduction to a section, and then provide either an example for every function (in which case you use the subheadings like we have) or your subheadings are focused on examples and use cases for the functions. Does that difference make sense? I'd love to see the latter (function list like Prom docs and then real-world examples and explanations), but it takes more time than providing the function heading, with a brief explanation, and then a simple example. I've captured this conversation in the metrics query doc issue. |
* added min_over_time traceql metric * fix parsing * add metricsAggregateMinOverTime to the extract conditions to fetch the attribute * fix tests * refactor code * refactor FloatizeAttribute * unused parameter warning * added a recombiner aggregator * ammend traceql lex * handle NaN values * fix tests * test that combination of valid values remove default NaNs * do not show empty time series * refactor aggregator function * changelog and documentation * doc changes * typo * bring back quantile doc header * fix typos
3 participants
What this PR does:
It adds a new TraceQL metric
min_over_timeused to compute the minimum value of a span series over a specified date range.Checklist
CHANGELOG.mdupdated - the order of entries should be[CHANGE],[FEATURE],[ENHANCEMENT],[BUGFIX]