Services & System Settings
Where the operational surfaces are
Four of them sit in the header, not inside Settings, because they are things you look at rather than things you change:
| Header entry | Shows |
|---|---|
| Status | Connection monitor, gateway and host information, CPU, memory, every disk QUBIQ writes to, pipeline statistics, and who is connected. |
| Logs | System logs and pipeline execution results, in two tabs. |
| Connections · Namespace | The connection list and the tag tree. |
Settings is where configuration lives, in fourteen categories:
| Category | Holds |
|---|---|
| System Health | Log retention, connection-monitoring mode, namespace optimisation |
| System Services | Start, stop and restart each service |
| Execution | Pipeline execution configuration |
| Storage Folders | Where QUBIQ keeps its data on disk |
| AI Assistant | Model, throughput, service identity, external access, embeddings, knowledge base |
| Map | Tile server and WMS sources |
| Alarms | Alarm storage |
| Audit Log | Audit storage, and the journal itself |
| Historian | Historian storage and concurrency |
| Store & Forward | Buffer status, buffer limits, per-buffer pause / rate / purge, and the quarantine |
| Security Policy | Sign-in settings, server certificate, external-content policy, security hardening, the permission editor |
| Identity & Access | Users, roles and on-call rosters |
| Script Sandbox | Sandbox level and installed Python packages |
| Backup & Restore | Take and restore backups |
Licensing is not in Settings: it has its own page at /licensing, reached from the licence status in the status bar. → Licensing & modules
Service control
Settings → System Services lists every service with its state and heartbeat.
| Action | Permission |
|---|---|
| View the service list | Access Settings |
| Start, stop or restart a service | Manage System Services |
Stopping through the interface tells the supervisor not to restart the service, so it stays down. That is the safe way to take a protocol worker out of service for maintenance — killing the process merely triggers an automatic restart.
For what each service does and what a restart affects, see The gateway.
Restarting a service belonging to a lapsed module asks you to confirm, naming that module. On a healthy installation the confirmation never appears — which is what stops it being trained away by the time it matters. → Licensing
Health and status
| Where | Shows | Permission |
|---|---|---|
| Status (header) | Host CPU, memory, every volume QUBIQ writes to, and gateway information | View Status Panel |
| Status (header) | Per-connection state and health | View Status Panel |
| Status (header) | Pipeline statistics | super admin |
| Status (header) | Connected users — who, from where, on what, and since when | View Status Panel |
Settings → System Services | Per-service state and heartbeat | Access Settings |
| The startup screen | What happened during boot | shown automatically |
The gateway also answers an unauthenticated liveness check suitable for a load balancer or container probe — ask your QUBIQ contact for the path if you need to wire one up.
Read the startup screen first after a failed or partial start. QUBIQ holds the interface behind it until every service reports in, and it records what the supervisor did and where it stopped — which is nearly always the answer.
The Connected column on the connected-users list carries a date as well as a time
New in 1.0.4. A LiveView screen on a wall stays up for days, and a bare 14:03 on such a row
reads as a few minutes ago when it was Tuesday. The full instant is in the cell's tooltip.
System settings
| Setting group | Permission |
|---|---|
| Log retention and system configuration | Manage Log Config |
| Connection-monitoring mode | Manage System Services |
| Storage folders | Manage System Services |
| Script sandbox | Manage Script Sandbox |
| Sign-in settings and security hardening | Manage Auth Settings |
| External-content policy | Access Settings / Manage Auth Settings |
| Map tile / WMS sources | Access Settings / Manage System Services |
| Historian storage and designation | View Connections / Edit Connection |
| Store & Forward: status / limits and buffer actions / quarantine import | View Connections / Edit Connection / Manage Backup & Restore |
| Alarm storage | Manage Alarms |
| Audit storage | Manage Audit |
| Users, roles and rosters | Manage Users |
| Backup and restore | Manage Backup & Restore |
External-content policy
Controls whether views may load content from outside the gateway — external images, embedded pages, remote map tiles. On an isolated plant network the answer should be no, and this is where you say so.
Map tiles
The OpenStreetMap widget can be pointed at a self-hosted tile server or a WMS endpoint, which is what makes geographic screens workable on a network with no internet access.
Storage
| What | Where |
|---|---|
| Configuration database | In the data directory, encrypted at rest |
| Project resources | projects/ |
| User SQLite databases | user-databases/ (sandboxed) |
| Python packages | python-packages/ |
| SNMP MIBs | mibs/ |
| Diagnostic reports | log/ |
| Per-service logs 1.0.8 | log/services/ |
| Store-and-forward buffers, quarantine, shared runtime state 1.0.8 | nats_data/ |
| Tag history | External QuestDB |
| Audit journal | External SQL database |
Choosing where they live New in 1.0.1
Settings → Storage Folders lets you browse for each location rather than typing a path blind.
The picker runs on the server, so it shows the gateway machine's filesystem — which is the one that
matters — and suggests the conventional location for the operating system it is running on.
Monitor free space on every volume QUBIQ writes to. The Status panel lists them all New in 1.0.1 rather than only the one the process happened to start from, which is the distinction that matters once history or project assets live on a separate disk.
The configuration database stays small; project assets and uploads are what grow — and, during a long outage of a destination, the store-and-forward buffers. The historian's buffer may use up to a quarter of the data volume by default, and if free space on that volume falls below its floor every buffer is frozen and a critical alarm is raised. 1.0.8 → The free-disk floor
A storage folder cannot hold a database's files New in 1.0.7
Saving an uploads or library folder that already contains another product's data files is refused, and the message names the product, where inside the folder it was found, and the file that gave it away — PostgreSQL, MongoDB, MySQL or MariaDB, QuestDB, Redis, InfluxDB and Elasticsearch are recognised, up to three levels down.
The case that prompted it: an uploads folder pointed at the parent of a server's Docker database volumes. Three things went wrong at once. Every media-inclusive backup archived live database files — which looked like a database backup and was not one — the File Manager offered rename and delete over them to any operator allowed to manage files, and walking hundreds of megabytes on the same disk the historian writes to made the whole machine sluggish.
An ordinary folder that already holds files is still accepted — that is how an existing uploads directory is carried to a new install. The check only refuses what it recognises, and if it cannot read the folder it lets the save through.
A folder listing in the File Manager shows at most 2,000 entries. New in 1.0.7 A larger folder says so — Showing the first 2,000 of 50,000 items — open a subfolder to see the rest — instead of stalling the request and the browser tab. The entries not shown still exist, still take space, and are still carried by a backup, which always walks the whole folder.
Time
Every timestamp an operator sees — chart axes, alarm journal entries, audit records — comes from the gateway's clock, not the browser's. Browsers align to it before they even have a session, so charts from two different panels agree.
Keep the host clock synchronised (NTP). Historian timestamps, alarm journal entries and audit records all come from the server. A drifting clock corrupts all three, and the damage is not retroactively fixable.
Diagnostics and reporting
| Where | What | Permission |
|---|---|---|
| Logs (header) → System | Read system logs | View System Logs |
| Logs (header) → System | Clear system logs | Clear System Logs |
| Logs (header) → Executions | Pipeline runs and per-node results | View Execution Logs |
| The report dialog | Submit a structured diagnostic report | authenticated |
| Status (header) | Pipeline statistics | super admin |
Diagnostic reports are written as rotated NDJSON under log/ in the data directory, and each
service's own output under log/services/. → Per-service logs
Routine operations
Daily — glance at connection health, active alarms, and free disk.
Weekly — review alarm bad actors, check store-and-forward buffers and quarantine, look at execution-log error rates.
Monthly — review users and permissions, verify the audit chain, confirm backups restore, review historian growth against retention.
Before every upgrade — take a backup, note the current version, read What's New.