Graph label issues: Group by format does not distinguish multiple values, alias does not support partitions

[bijwaard]

Dear all,

I found two ways to change the field label in the Grafana legend with the tdengine plugin:

• using alias which supports naming fields separated by comma's and overrides "group by format"
• using "group by format" which currently doesn't seem to display the field names (they all get the same name in the legend)

It would be great if the group by format would also support using field names, either by reusing the field-name in the query or by explicitly giving each field a name like in the alias. Or alternatively, that I can use the group format result inside the alias.

I have queries on my supertable rt_table like:
select _wstart as ts,avg(v0) as L1,avg(v1) as L2,avg(v2) as L3,sensor,signal,type from lvsensors.rt_table where sensor='${SelectedSensor}' and signal='power_median' and type in '${SelectedType} and ts > $from and ts < $to partition by sensor,signal,type interval($interval) fill(null);
And would like to use the field-names L1, L2 and L3 in the legend as well as some of the partition names, so that I can create an alias like <field-name>-{{type}}, where {{type}} is either active or reactive using the multi-select variable $SelectedType.

I currently have separate queries for each $SelectedType with different alias (L1-active,L2-active,L3-active respectively L1-reactive,L2-reactive,L3-reactive), but cannot multi-select the type to (un)display the them. I would like to do the same with $SelectedSensor to enable showing all sensor graphs in one timeseries panel instead of the repeating rows I have now with legend name {{sensor}}:<field-name>-{{type}} and the following query:

select _wstart as ts,avg(v0) as L1,avg(v1) as L2,avg(v2) as L3,sensor,signal,type from lvsensors.rt_table where sensor in '${SelectedSensor}' and signal='power_median' and type in '${SelectedType} and ts > $from and ts < $to partition by sensor,signal,type interval($interval) fill(null);

P.s. Thanks for this great plugin, I'm trying to use tdengine as alternative to influxdb and noticed big performance gains for memory as well as processing on the server- and client-side. Still have to see how the performance is after moving multiple-years of data form influxdb to tdengine.

Kind regards,
Dennis

[bijwaard]

To illustrate the issue another way, the value label names L1, L2 and L3 are not shown when the group by column is used.

The data from the query looks like this:

taos> select ts,v0 as L1,v1 as L2,v2 as L3, sensor,signal,type from lvsensors.rt_table where signal='power_median' and ts>1751453145000000 and ts>1751459950000000 and ts<1751459952000000 partition by sensor,signal,type;
             ts             |            l1             |            l2             |            l3             |             sensor             |             signal             |              type              |
====================================================================================================================================================================================================================
 2025-07-02 14:39:11.000000 |              -1223.507971 |                -395.04186 |              -2592.380466 | VT                             | power_median                   | active                         |
 2025-07-02 14:39:11.000000 |              -1223.507971 |                -395.04186 |              -2592.380466 | CT                             | power_median                   | active                         |
 2025-07-02 14:39:11.000000 |         -224.750417237027 |         -110.648156282834 |         -352.576829065648 | CT                             | power_median                   | reactive                       |
 2025-07-02 14:39:11.000000 |              -1356.741086 |               -1263.08277 |              -1283.752571 | CT2                            | power_median                   | active                         |
 2025-07-02 14:39:11.000000 |         -3.58487173824727 |          33.4294043741199 |          107.689723149007 | CT2                            | power_median                   | reactive                       |
 2025-07-02 14:39:11.000000 |         -224.750417237027 |         -110.648156282834 |         -352.576829065648 | VT                             | power_median                   | reactive                       |
Query OK, 6 row(s) in set (0.006291s)

When I group columns on sensor, signal,type with {{sensor}}:{{signal}}-{{type}} the line (L1, L2, L3) is not shown in the label, when I use the alias, I have to write separate queries for all the different group permutations (sensor, signal, type).
It would be great if we could also use the value-names as a group.

I.e. the value-name in the column format {{sensor}}:{{value}}-{{signal}}-{{type}} is not shown:

[bijwaard]

I can now display the graphs in one panel, but need to do separate queries per value (phases L1, L2, and L3), which is probably less efficient.