How to merge rows from 3 Loki queries by 1-second time buckets in Geomap?

[toon2]

• What Grafana version and what operating system are you using?

Grafana Cloud v12.4.0

• What are you trying to achieve?

I have a Geomap panel with 3 Loki queries that each return different data with their own timestamps:

• Query A: GPS coordinates (latitude, longitude)

• Query B: “No image” event duration (ms)

• Query C: Reconnect event duration (ms)

I want to merge these into a single row per time bucket so I can plot markers on the map with color/size based on duration.

• How are you trying to achieve it?

Since the timestamps don’t match exactly across queries (millisecond precision), I created a calculated field Time / 1000 to bucket into 1-second intervals, then used:

1. Extract fields — parse lat, lon, duration from JSON log lines

2. Convert field type — to numbers

3. Calculate field — Time / 1000 for 1s bucketing

4. Join by field — outer join on Time / 1000

5. Organize fields — rename/hide duplicates

6. Group by — group on Time / 1000 with firstNotNull for coordinates, max for durations

• What happened?

Rows with the same Time / 1000 value do not merge. For example, timestamp 1770642598 shows up as 3 separate rows instead of one:

Time / 1000	latitude	longitude	no-image-duration	reconnect-duration
1770642598			4501
1770642598				534
1770642598	51.0	3.84

• What did you expect to happen?

All three rows to collapse into one:

Time / 1000	latitude	longitude	no-image-duration	reconnect-duration
1770642598	51.0	3.84	4501	534

• Can you copy/paste the configuration(s) that you are having problems with?

No errors in the UI. My suspicion is that Time / 1000 produces slightly different floating-point values per query (e.g. 1770642598.123 vs 1770642598.456), so the Group By treats them as distinct keys even though they display identically.

[infofcc3]

Please try this

[toon2]

Hi Shanker,

Thanks for the answer, it worked.

I added a join by field (Outer(time series)) and organize field by name between the Binary and Unary operation. The binary operation creates multiple Time/1000 fields (one for each tab listed at start before the join).

solution: use unary operation floor operation to round the values (not visible) and then Group by operation and do the Group by on that rounded off field.

EDIT: Do not forget to do *1000 for the time field and convert this back to a time format with Convert Field Type transformation

Overview

[infofcc3]

Thanks @toon2 for your feedback