Grafana is an open source volume metrics monitoring and visualization tool, which can be used to present time series data and analyze application runtime status.

We developed the Grafana-Plugin for IoTDB, using the IoTDB REST service to present time series data and providing many visualization methods for time series data.

This plugin requires Grafana 12.3.0 or later. Earlier versions, including the 9.x, 10.x and 11.x releases supported by previous plugin releases, are no longer supported.

The floor is not arbitrary. The @grafana/* packages this plugin builds against are webpack externals: they are resolved from the host Grafana at runtime rather than bundled into the plugin. A plugin built against a newer Grafana therefore compiles cleanly and then fails inside an older host, with no bundled fallback — so a declared floor is only meaningful if something actually boots that version and loads the plugin there.

CI does exactly that. The minimum-grafana-version job parses the floor out of plugin.json, boots that Grafana release together with IoTDB, and checks that the plugin's frontend loads, that a table-model query returns data through the native client, and that the table-model editor lists databases, tables and columns from the server. Raising or lowering the declared floor therefore changes what is tested, with no separate file to keep in step.

Grafana displays data in a web page dashboard. Please open your browser and visit http://<ip>:<port> when using it.

• IP is the IP of the server where your Grafana is located, and Port is the running port of Grafana (default 3000).

• The default login username and password are both admin.

Click the Settings icon on the left, select the Data Source option, and then click Add data source.

Select the Apache IoTDB data source.

• Fill in http://<ip>:<port> in the URL field
  • ip is the host ip where your IoTDB server is located
  • port is the running port of the REST service (default 18080).
• Enter the username and password of the IoTDB server

Click Save & Test, and Success will appear.

Click the Dashboards icon on the left, and select Manage option.

Click the New Dashboard icon on the top right, and select Add an empty panel option.

Grafana plugin supports SQL: Full Customized mode and SQL: Drop-down List mode for the tree model, and SQL: Table Model mode for the table model. The default mode is SQL: Full Customized mode.

Enter content in the SELECT, FROM , WHERE and CONTROL input box, where the WHERE and CONTROL input boxes are optional.

If a query involves multiple expressions, we can click + on the right side of the SELECT input box to add expressions in the SELECT clause, or click + on the right side of the FROM input box to add a path prefix:

SELECT input box: contents can be the time series suffix, function, udf, arithmetic expression, or nested expressions. You can also use the as clause to rename the result.

Here are some examples of valid SELECT content:

• s1
• top_k(s1, 'k'='1') as top
• sin(s1) + cos(s1 + s2)
• udf(s1) as "alias"

FROM input box: contents must be the prefix path of the time series, such as root.sg.d.

WHERE input box: contents should be the filter condition of the query, such as time > 0 or s1 < 1024 and s2 > 1024.

CONTROL input box: contents should be a special clause that controls the query type and output format. The GROUP BY input box supports the use of grafana's global variables to obtain the current time interval changes $__from (start time), $__to (end time)

Here are some examples of valid CONTROL content:

• GROUP BY ([$__from, $__to), 1d)
• GROUP BY ([$__from, $__to),3h,1d)
• GROUP BY ([2017-11-01T00:00:00, 2017-11-07T23:00:00), 1d)
• GROUP BY ([2017-11-01 00:00:00, 2017-11-07 23:00:00), 3h, 1d)
• GROUP BY ([$__from, $__to), 1m) FILL (PREVIOUSUNTILLAST)
• GROUP BY ([2017-11-07T23:50:00, 2017-11-07T23:59:00), 1m) FILL (PREVIOUSUNTILLAST)
• GROUP BY ([2017-11-07T23:50:00, 2017-11-07T23:59:00), 1m) FILL (PREVIOUS, 1m)
• GROUP BY ([2017-11-07T23:50:00, 2017-11-07T23:59:00), 1m) FILL (LINEAR, 5m, 5m)
• GROUP BY ((2017-11-01T00:00:00, 2017-11-07T23:00:00], 1d), LEVEL=1
• GROUP BY ([0, 20), 2ms, 3ms), LEVEL=1

Tip: Statements like select * from root.xx.** are not recommended because those statements may cause OOM.

Select a time series in the TIME-SERIES selection box, select a function in the FUNCTION option, and enter the contents in the SAMPLING INTERVAL、SLIDING STEP、LEVEL、FILL input boxes, where TIME-SERIES is a required item and the rest are non required items.

Queries IoTDB's table model (IoTDB 2.x) through the native Go client (apache/iotdb-client-go). Enter a standard SQL statement in the SQL input box, for example:

SELECT time, device_id, temperature FROM table1 WHERE $__timeFilter(time)

Unlike the tree-model modes, which go through the REST service, this mode connects to the IoTDB RPC port (6667 by default). The data source's rpc address option sets that endpoint explicitly; when it is left empty, the URL's host with port 6667 is used.

DATABASE: the database to run the statement against, picked from the databases the server lists (a name can also be typed, for example a $database template variable). Required — every query first runs USE <database> on its pooled session, so results stay deterministic regardless of which session the pool hands out. Fully-qualified database.table references are still allowed and take precedence over the session database.

TABLE (optional): picking a table lists its columns grouped by category (TAG, FIELD, ATTRIBUTE). Use starter query then replaces the SQL with a query that selects the table's time, TAG and numeric FIELD columns (every FIELD when none is numeric) within the dashboard time range, for example SELECT "time", "region", "temperature" FROM "weather" WHERE $__timeFilter("time"). Names are always double-quoted, so reserved words such as order and names with spaces stay valid. The table only drives this helper; the SQL is what runs.

SQL input box: a single table-model SELECT statement. The following macros are expanded by the plugin before the statement is sent:

The macros expand to ISO 8601 timestamp literals (e.g. 2020-09-13T12:26:40.000+00:00), which the server interprets in its own configured timestamp_precision — so time filtering works unchanged on ms, us and ns servers, and TIMESTAMP values are likewise converted by the client using the server-reported precision.

QUERY TYPE option:

• Range (default): evaluates the SQL over the complete Grafana time range. Queries saved before this option existed remain Range queries.
• Instant: evaluates the SQL over the complete Grafana range, then returns the latest row at or before the range end for each series, with its timestamp normalized to that evaluation time. This supports samples whose timestamps do not exactly equal the panel end. SQL that needs a window boundary aligned to the evaluation time (for example HOP(... ORIGIN => ...)) must explicitly use $__evaluationTime; the plugin does not rewrite ORIGIN values, comments, or string literals.
• Both: executes Range and Instant independently and returns both results in distinctly named frames.

FORMAT option:

• Time series (default): rows are sorted by the first TIMESTAMP column, and a result that contains tag/string columns next to numeric columns is pivoted into one series per tag combination — so a query returning several devices renders as separate lines in a time-series panel. Note that the pivot requires timestamps without nulls; results are sorted by the plugin, so an ORDER BY clause is not needed.
• Table: rows are returned exactly as the server sent them (use this with the table panel, or when your ORDER BY should be preserved).

Result columns are typed from the result set's data types: TIMESTAMP columns become Grafana time fields, INT32/INT64 become integers, FLOAT/DOUBLE become floats, BOOLEAN becomes booleans, and TEXT/STRING/DATE/BLOB render as strings.

There is no server-imposed row cap on this path; add a LIMIT clause when querying wide time ranges over dense data to bound the result size.

Both SQL: Full Customized and SQL: Drop-down List input methods support the variable and template functions of grafana. In the following example, raw input method is used, and aggregation is similar.

When a multi-value template variable is used in the FROM input box (prefixPath), the plugin automatically expands it into multiple paths. This enables the common "global variable filter" pattern where a single dashboard dropdown controls all panels.

For example, define a multi-select variable device with values device1, device2, ..., device8. Then in the FROM input box, enter:

root.application.${device}

When the user selects device1 and device2, the plugin internally expands this to:

root.application.device1
root.application.device2

Only the selected devices are queried from IoTDB — no client-side filtering or transformations needed.

This works with any number of prefixPath entries. Literal paths (without variables) and single-value variables behave the same as before.

After creating a new Panel, click the Settings button in the upper right corner:

Select Variables, click Add variable:

Example 1：Enter Name, Label, and Query, and then click the Update button:

Apply Variables, enter the variable in the grafana panel and click the save button:

Example 2: Nested use of variables:

Example 3: using function variables

The Name in the above figure is the variable name and the variable name we will use in the panel in the future. Label is the display name of the variable. If it is empty, the variable of Name will be displayed. Otherwise, the name of the Label will be displayed. There are Query, Custom, Text box, Constant, DataSource, Interval, Ad hoc filters, etc. in the Type drop-down, all of which can be used in IoTDB's Grafana Plugin For a more detailed introduction to usage, please check the official manual (https://grafana.com/docs/grafana/latest/variables/)

In addition to the examples above, the following statements are supported:

• show databases
• show timeseries
• show child nodes
• show all ttl
• show latest timeseries
• show devices
• select xx from root.xxx limit xx 等sql 查询

Tip: If the query field contains Boolean data, the result value will be converted to 1 by true and 0 by false.

This plugin supports Grafana alert function.

For more details about Grafana operation, please refer to the official Grafana documentation:https://grafana.com/docs/grafana/latest/alerting/