influxWeb

A web-based admin UI for a local InfluxDB v2 instance. Built for editing smarthome telemetry (e.g. from ioBroker) without fighting the influx CLI: browse buckets, filter datapoints by measurement/tag, inspect and edit values, fix timestamps, delete ranges, and export/import data as ODS spreadsheets (for editing in LibreOffice Calc with correct types, no CSV locale/decimal-separator guessing).

Runs as a small FastAPI backend + a static HTML/JS frontend (Tabulator.js), meant for LAN-only deployment on something like a Raspberry Pi — no login, no internet exposure.

This project was developed against an ioBroker InfluxDB-history setup specifically (e.g. its ack/from/q tag conventions show up in a couple of default-value choices) — it should work against any InfluxDB v2 instance, but has not yet been tested outside that one environment.

ioBroker's InfluxDB adapter can store metadata in two different ways, depending on the "store metadata (ack, from, q) as tags" setting in its Expert config:

• Tag-based (setting enabled): ack, from, q are InfluxDB tags, value is the field.
• Field-based (setting disabled): ack, from, q, and value are all InfluxDB fields, no tags.

Both variants are fully supported when INFLUXWEB_MODE=iobroker is set in .env (pre-configured in .env.example). influxWeb auto-detects the storage variant per measurement, groups field-based data into one row per logical point, and handles Add point, Edit, Delete, Retime, and ODS Export/Import correctly for both variants. The "Group fields by point" toggle is hidden in iobroker mode (grouping is always on).

Set INFLUXWEB_MODE=default in .env to disable ioBroker-specific behaviour. In that mode, all operations work for tag-based data. Field-based data can be browsed, queried, exported, edited, retimed, and deleted correctly — but "Add point" will not create a complete ioBroker point and should be avoided.

Features

• Bucket selection
• Measurement/tag schema browsing and selection, file-explorer-style (click, Ctrl+click, Shift+click), with a clear/reset action and a text filter
• Time range selection
• Querying and listing datapoints, with adjustable page size and multi-row selection (same click/Ctrl/Shift model as the schema tree); capped at 200,000 points per query to keep memory use bounded regardless of how broad a selection is — narrow the measurement/tag/time filter if a result comes back marked as truncated
• A Statistical View toggle: instead of one row per point, one summary row per measurement+field (count, and for numeric fields min/max/mean/standard deviation) over the currently loaded query result
• A "Group fields by point" toggle: pivots the Data View from one row per field to one row per logical reading (fields as columns) - mainly useful for ioBroker buckets using field-based storage, where each point otherwise shows as several separate rows. Off by default, remembered across visits. Hidden when INFLUXWEB_MODE=iobroker is active (grouping is always on there)
• Export to ODS (whole query result, or just the selected rows), with proper cell types (numbers, booleans, dates) and an instructions block describing the round-trip contract for a later import
• Import from ODS: re-upload a previously exported (and possibly hand-edited) file. A dry-run preview shows what would be written, which bucket(s) the file targets, and any rows that will be skipped (with a reason) before you confirm. Edited rows overwrite the matching point; new rows create one; deleted rows are left untouched in InfluxDB (import only writes, never deletes)
• Export Raw / Import Raw: a deliberately bare-bones pair for CLI/scripting interop, using InfluxDB's own annotated CSV format (influx query/ influx write --format csv). Export Raw dumps the entire active bucket for the current time range in one query, no measurement/tag filtering. Import Raw writes a single-schema CSV file straight into the active bucket with no preview - it stops at the first row it can't parse or write, rather than trying to validate the whole file upfront
• Delete points - either exactly the selected rows, or (with nothing selected) every row currently loaded in the table - with a preview and explicit confirmation before anything is deleted
• Edit a value inline (double-click a cell in the Value column) with a confirm-before-save step, or add a brand-new point via a separate modal — both share the same write path (overwrite by measurement+tags+timestamp, last-write-wins, same as InfluxDB itself). In INFLUXWEB_MODE=iobroker, the metadata fields ack, from, and q are also editable inline: for field-based storage they behave like any other field edit; for tag-based storage the confirm step writes a new point with the changed tag and deletes the old one. q uses a dropdown of the known ioBroker quality-flag values (0 Good, 1 General error, 2 No connection, 16 Substitute, 32 Device error, 64 Device-specific error, 128 Not connected)
• Retime points: shift the timestamp of selected (or all loaded) points by a calendar-aware offset, or normalize them to the start of the minute/hour/day/week/month/year (local time, DST-aware) — both via the "Retime in range" button with the usual preview/confirm step. A single point's timestamp can also be edited in place (double-click a cell in the Time column)

Sorting

Column headers use Tabulator's standard click behavior, not anything custom to influxWeb:

• Click a header to sort by that column alone.
• Shift+click another header to add it as a further sort level. The most-recently Shift-clicked column becomes the new primary sort key, and any columns sorted before it become tiebreakers underneath it.

This takes a moment to get used to, but covers the common case of wanting to read one measurement's values in chronological order without any extra UI:

1. Click Time to sort by it.
2. Shift+click Measurement. Measurement becomes the primary sort, with Time still applied underneath as the tiebreaker - so within each measurement, rows come out time-ordered.

Setup

Installs influxWeb as a systemd service on a Raspberry Pi (or any Linux host with systemd), running under its own dedicated system user, into /opt/influxweb.

Quick install

curl -sLf https://raw.githubusercontent.com/MyHomeMyData/influxWeb/main/deploy/install.sh | sudo bash -

This clones the repo into /opt/influxweb, creates the dedicated influxweb system user, sets up the virtual environment, installs the systemd unit, and installs the influxweb-upgrade/influxweb-uninstall commands (see Upgrading and Uninstalling) — it does not start the service yet, since InfluxDB access still needs to be configured. After it finishes:

1. Fill in InfluxDB access:

   sudo -u influxweb nano /opt/influxweb/.env   # fill in INFLUX_URL / INFLUX_TOKEN / INFLUX_ORG

   The .env.example pre-sets INFLUXWEB_MODE=iobroker. Set it to default to disable ioBroker-specific behaviour (see above for details).
2. Start the service:

   sudo systemctl enable --now influxweb

3. Open http://<pi-host>:8085/ from another machine on your LAN.

Manual installation

Equivalent step-by-step version of the above, if you'd rather not pipe a script into sudo bash, or want to adjust something along the way.

1. Clone the repository:

   sudo git clone https://github.com/MyHomeMyData/influxWeb.git /opt/influxweb

2. Create a dedicated system user (matches User=/Group= in deploy/influxweb.service) and hand it ownership of the install. --home-dir points it at /opt/influxweb itself, since that's the only directory it owns - without it, tools like pip fall back to a non-existent default home and disable their cache with a permissions warning:

   sudo useradd --system --no-create-home --home-dir /opt/influxweb --shell /usr/sbin/nologin influxweb
   sudo chown -R influxweb:influxweb /opt/influxweb

3. Create the virtual environment and install dependencies:

   sudo -u influxweb python3 -m venv /opt/influxweb/.venv
   sudo -u influxweb /opt/influxweb/.venv/bin/pip install -r /opt/influxweb/requirements.txt

4. Configure InfluxDB access:

   sudo -u influxweb cp /opt/influxweb/.env.example /opt/influxweb/.env
   sudo -u influxweb nano /opt/influxweb/.env   # fill in INFLUX_URL / INFLUX_TOKEN / INFLUX_ORG

   .env is gitignored — never commit it.
5. Install and start the systemd service:

   sudo cp /opt/influxweb/deploy/influxweb.service /etc/systemd/system/
   sudo systemctl daemon-reload
   sudo systemctl enable --now influxweb

6. Open http://<pi-host>:8085/ from another machine on your LAN.

influxWeb binds 0.0.0.0 by default — keep it reachable only from your LAN (no router port-forward); there is no authentication built in.

Upgrading

sudo influxweb-upgrade

Installed automatically by Quick install above — pulls the latest code, updates dependencies, and restarts the service. If you used the manual installation instead, do the same steps by hand:

cd /opt/influxweb
sudo -u influxweb git pull
sudo -u influxweb .venv/bin/pip install -r requirements.txt
sudo systemctl restart influxweb

Check the Changelog below for anything version-specific to be aware of before upgrading. The current running version is shown in the page header.

Uninstalling

sudo influxweb-uninstall

Installed automatically by Quick install above. After a confirmation prompt, stops and removes the service, /opt/influxweb (including .env), the influxweb system user, and the influxweb-upgrade/influxweb-uninstall commands themselves. If you used the manual installation, you're on your own for cleanup — stop/disable the service, then remove the systemd unit, /opt/influxweb, and the system user.

Temporary setup for testing

For trying out influxWeb without installing it as a service:

python3 -m venv .venv
.venv/bin/pip install -r requirements.txt
cp .env.example .env   # then fill in INFLUX_URL / INFLUX_TOKEN / INFLUX_ORG
.venv/bin/uvicorn app.main:app --reload --host 0.0.0.0 --port 8085

Open http://localhost:8085/ (or http://<host>:8085/ from another machine on the LAN - uvicorn only listens on 127.0.0.1 unless --host 0.0.0.0 is passed explicitly).

Changelog

0.3.1 (2026-07-03)

(MyHomeMyData) In INFLUXWEB_MODE=iobroker, the metadata fields ack, from, and q are now editable inline in the data table:

• Field-based storage: ack/from/q appear as editable field columns (instead of read-only synthetic tag columns); q uses a dropdown of the eight known ioBroker quality-flag values; unknown values appear as an extra option
• Tag-based storage: double-clicking ack, from, or q opens a confirm dialog that writes a new point with the changed tag and deletes the old one (same write-before-delete pattern as Retime)

Also fixed: q was displayed as a decimal (0.0) in field-based mode — now shown as an integer (0).

0.3.0 (2026-07-01)

(MyHomeMyData) Added INFLUXWEB_MODE=iobroker for transparent, full-featured support of ioBroker's InfluxDB adapter — covering both its tag-based and field-based storage variants simultaneously, auto-detected per measurement at query time.

In iobroker mode, field-based data is grouped into one row per logical point on the backend (measurement + timestamp → one display row with ack/from/q shown as synthetic tag columns). All operations work correctly for both storage variants:

• Add point: dedicated form with ack/from/q fields; writes four single-field calls that InfluxDB merges into one complete point
• Inline edit: writes to the correct tagless series (not a new tagged one)
• Delete / Retime: use empty tags for field-based points, correctly identifying the series by measurement and timestamp alone
• ODS Export: one row per point; ack/from/q stored in extra.* columns (e.g. extra.ack, extra.ack_type) alongside the main value — round-trip safe
• ODS Import: detects extra.* columns and writes each as a separate field to the same series+timestamp; InfluxDB merges them into one complete point. Also fixed LibreOffice re-saving boolean cells as numeric (1/0) breaking import

The "Group fields by point" toggle is hidden in iobroker mode (grouping is always active). In default mode, all existing behavior is unchanged.

Also fixed in this release: number-input stepper buttons (▲▼) in the inline cell editor triggered the confirm modal immediately on click instead of waiting for Enter or blur.

0.2.2 (2026-06-30)

(MyHomeMyData) Fixed several point-count inconsistencies introduced by 0.2.0's "Group fields by point" toggle: the toolbar buttons, Retime/Delete dialogs, and the points-loaded status line could show the raw per-field row count instead of the logical point count, and toggling grouping didn't refresh the status line (in one case leaving "Querying..." stuck after a query that legitimately returned zero points). Also fixed Delete sending one redundant delete call per field of a point instead of one per point.

0.2.1 (2026-06-30)

(MyHomeMyData) Added a warning in the Retime confirmation dialog when a hand-edited timestamp introduces sub-millisecond precision that wasn't there before, since this is usually an accidental stray edit of the trailing digits rather than intentional (found by Marc Berg while testing 0.2.0).

0.2.0 (2026-06-30)

(Marc-Berg) Added a "Group fields by point" Data View toggle, pivoting one row per field into one row per logical reading - mainly useful for ioBroker buckets using field-based storage.

0.1.1 (2026-06-28)

(MyHomeMyData) Fixed an unfiltered selection querying the entire bucket, and incorrect row order for ioBroker's field-based storage mode. Documented current support boundaries between ioBroker's tag-based and field-based storage modes.

0.1.0 (2026-06-24)

(MyHomeMyData) Initial beta release.

License

MIT, see LICENSE.