Knowledge baseOperate
Backups, recovery and updates
The nightly backup on your server and how to check it, incremental backups for large databases, restoring, the system health card, data retention, and updates from the interface.
On your server, backups and updates are yours to run — SyncTide does most of it by itself and tells you when something is wrong. In our cloud we do both; see the end of this page.
Your measurement history lives in PostgreSQL, not in the SyncTide folder. That has one reassuring consequence — uninstalling SyncTide while keeping the data leaves the database where it is — and one dangerous one: removing PostgreSQL removes everything. Keep a copy of the backups off the server.
The nightly backup
The installer registers a Windows scheduled task, SyncTide Backup, every night at 02:00, running as SYSTEM so it works with nobody signed in. It runs backup_platform.ps1 from the installation folder and writes one .zip per night to the backups\ folder there.
- Each archive holds a full database dump plus the settings (
.env, the runtime configuration), the logs, the report templates, the PDF header logos and the intervention photos. New in 1.14.8 — coming soon - The dump is read back before the archive is kept. A run that cannot vouch for its archive stops with the reason; a run that does not end with
Created archive:produced no backup. - The newest 7 archives are kept, and a run refuses to start when the disk cannot hold another one. New in 1.14.7 — coming soon
From the interface
In Configuration → Runtime → Retention, the Backups card lists the archives with their date and size, downloads any of them and takes one on demand. Take one before every upgrade.
Copy a backup off the server. A backup on the same disk as the database protects you from mistakes, not from the disk failing. Download an archive now and then, or copy the
backups\folder to another disk or a NAS.
By hand
From an elevated PowerShell in the installation folder (usually C:\Program Files\SyncTide):
| Command | Use for |
|---|---|
.\backup_platform.ps1 |
A normal backup: database plus settings |
.\backup_platform.ps1 -SupportBundleOnly |
Logs and settings to send to support, no data |
.\backup_platform.ps1 -SkipDatabase |
Settings only |
If the scheduled task is missing (group policy can block it), the installation still completes; register it again from an elevated prompt with schtasks, as described in BACKUP_AND_RESTORE.md in the installation folder, and check that an archive appears the next morning.
Large databases: incremental backups
New in 1.14.7 — coming soon
From 100 GB the same nightly task switches to incremental backups (PostgreSQL 17 or later): the first night copies the database server once, and every other night copies only the blocks that changed — new data, but also a value edited months back, a late datalogger file or a settings change.
- Every 7 nights the chain is combined, on the backup disk, into a new full copy, so the database is never read in full again.
- The newest two chains are kept: 8 to 14 nightly restore points, each verified before older ones are removed.
- The nightly
.zipstill holds the settings and files, and names the database point it goes with.
It switches only when PostgreSQL holds no other software’s databases — a physical backup is the whole PostgreSQL server. Otherwise the nightly dump continues, with a warning in the log.
Checking a backup without restoring it
You do not have to trust a backup you have never opened. Unzip an archive and list the dump’s contents; nothing is changed:
& "C:\Program Files\PostgreSQL\17\bin\pg_restore.exe" -l synctide.backup | Select-Object -First 20
Empty output or an error means the archive is unusable — better to find out today than during an incident.
Restoring
A nightly dump. TimescaleDB needs two extra steps around the restore; skipping them is the usual reason a restore of a good backup appears to fail.
- Stop SyncTide:
Stop-Service SyncTide* - Unzip the archive and find
synctide.backup. - Create an empty database (or drop the existing one — irreversible, be sure you have the backup you think you have).
- Restore with the TimescaleDB helpers around it:
$bin = "C:\Program Files\PostgreSQL\17\bin"
& "$bin\psql.exe" -U postgres -c "CREATE DATABASE synctide;"
& "$bin\psql.exe" -U postgres -d synctide -c "CREATE EXTENSION IF NOT EXISTS timescaledb;"
& "$bin\psql.exe" -U postgres -d synctide -c "SELECT timescaledb_pre_restore();"
& "$bin\pg_restore.exe" -U postgres -d synctide --no-owner synctide.backup
& "$bin\psql.exe" -U postgres -d synctide -c "SELECT timescaledb_post_restore();"
- If you are rebuilding the machine, put back
.envfrom the archive. ThenStart-Service SyncTide*, open the interface and check that the most recent values are there.
Warnings about circular foreign keys on continuous_agg during pg_restore are normal for TimescaleDB.
An incremental restore point New in 1.14.7 — coming soon, as an administrator on the server:
cd "C:\Program Files\SyncTide"
.\restore_platform.ps1 -List
.\restore_platform.ps1 -Point "<the point's folder from the list>"
It rebuilds the point and checks it, stops SyncTide and PostgreSQL, keeps the current data folder aside as <data>.pre-restore-<time>, puts the rebuilt one in its place and starts everything again — and puts the previous data back if PostgreSQL does not start. Delete the .pre-restore- folder once SyncTide looks right. It needs free space on the PostgreSQL drive of about the size of the database, and restores only on the same PostgreSQL major version.
The system health card
The System health card (same screen) answers “is this installation working right now?”, refreshed every minute, naming problems rather than summarising them:
| Check | What it means |
|---|---|
| Devices reporting | Each device against its own communication timeout — one silent datalogger cannot hide behind the others |
| Workers alive | Each background service is running and still doing work |
| Last backup | Age of the newest archive, flagged when overdue |
| Disk free | Free space on the volume of the database and the backups — they share it, so filling it stops both |
The health monitor also warns when that disk is nearly full: 90% used or under 2 GB free. New in 1.14.7 — coming soon
Data retention
The retention policy on the same screen runs a nightly purge of old time-series data, message history and ingested files, each with its own age limit and on/off switch; a pass can also be run at once to see its effect. Forecasts over communication gaps are kept 180 days by default, so a gap can still be filled weeks later.
Set retention longer than your backup rotation. Purged data that was never captured in a kept archive is gone.
Updates
When a newer signed release exists, administrators see a notice with four choices: Install, Install at 02:00, Remind me in 3 h and Don’t remind me. Both installs ask for a second confirmation.
- The download is checked against SyncTide’s release signature before anything runs.
- A backup is taken, the services stop, the installer upgrades in place and the services come back.
- Progress and the result are shown in Configuration → Runtime → Updates, which also shows the installed version and cancels a scheduled install.
A release newer than the end of your licence’s update period is not offered. If one is installed anyway, the licence is refused with a clear message and no data is lost: upload a renewed licence, or reinstall a version from within the period.
Without internet access on the server, run the newer installer instead: it finds the installation and upgrades it in place, with no wizard pages.
In our cloud
Every day the database and the files of each instance are backed up with an integrity check, kept for 30 days with the same provider in the EU; restoring one instance on its own is tested. Updates are applied by us — there is nothing to install, and a restart can interrupt the web interface for a few minutes.