Jobs
A job tells one agent what to back up, where to store it, and when to run. Jobs can back up a filesystem path or a PostgreSQL database. They can also call optional local hooks before and after a run.
Create a job
- Open Jobs and create a job. Name the job and select the agent that can access the source and destination.
- Choose Files and folders or PostgreSQL, then enter the source settings.
- Select a destination repository and schedule. See Destinations for repository URL formats and credentials.
- Optionally select a pre-hook or post-hook registered on that agent.
- Save the job and run the setup command shown by Vecta on the agent:
bash
sudo vecta-agent setup <JOB_ID>
Setup stores destination credentials locally and prepares the Restic repository when needed. PostgreSQL jobs also prompt for the database password. Neither kind of password is entered in the dashboard.
The same builder is used during first-run onboarding. For PostgreSQL tool requirements, local password setup, and hook registration, see Database Backups and Hooks.
Source types
Filesystem
Enter a path that exists on the agent machine. The agent must be able to read it. For an SFTP destination, enter an SSH port or leave it blank to use port 22.
To skip files or directories under that source, enter Exclude patterns, one Restic pattern per line. For example:
*.log
node_modules/*
cache/*
Patterns are passed to Restic as arguments; they are not expanded by a shell. Enter up to 100 patterns, each no longer than 256 characters. Excludes apply only to filesystem jobs. Leave the field empty to back up everything under the selected source.
PostgreSQL
Choose an agent that reports PostgreSQL backup support, then enter the database host, port, database name, username, and SSL mode. The database can run on another machine, provided the agent can reach it. Vecta stores these connection settings in the job, but never asks for or stores the database password in the dashboard. Setup on the agent stores that password locally.
Vecta streams a PostgreSQL custom-format archive into Restic instead of first writing a complete dump to a temporary file. Restic's cache, upload temporary files, or a local repository still need disk space. Compression and deduplication results vary; Vecta does not promise a fixed storage saving.
Snapshot retention
Retention is optional and configured per job in the Snapshot retention section. On the current Free plan, choose an age window from 1 to 14 days. After a successful backup, the agent applies the policy and reports snapshots it pruned. Calendar-based keep rules are not available on the Free plan.
Turning retention off stops automatic pruning for future runs; it does not bring back snapshots that were already pruned or remove older snapshots by itself. Review your retention window against your recovery needs before enabling it. Vecta's snapshots page marks reported pruned snapshots as unavailable for restore.
Hooks
Hooks are optional tasks that an administrator has registered on the selected agent. A pre-hook runs before source capture; a post-hook runs afterward. The job selects a hook by ID and may fill in only the parameters declared for it. It cannot define a command or executable path. Hooks granted root access are labelled in the builder; review what each hook does before making it available to jobs. See Database Backups and Hooks for setup and security details.
Vecta prepares the Restic repository before starting the visible pipeline. If a pre-hook fails, the source is not captured. After the pipeline starts, the agent attempts a configured post-hook even if capture or Restic fails or the run is cancelled. A post-hook failure makes the run fail, but the log keeps the ID of any snapshot that was already created.
Schedules
| Choice | Meaning |
|---|---|
| Every 6 hours | The default interval. |
| Every 12 hours | Runs twice per day. |
| Every day at 03:00 UTC | Runs daily at 03:00 UTC. |
| Weekly on Sunday at 02:00 UTC | Runs weekly at 02:00 UTC. |
| Custom cron | A validated five-field UTC cron expression. |
Warning
Schedules use UTC. Convert your local time before saving. The agent checks in every couple of minutes, so a job normally starts at the first check-in after it becomes due.
An interval and a cron schedule are mutually exclusive. Editing a schedule takes effect on the agent's next check-in.
Runs and controls
| Status or badge | Meaning |
|---|---|
pending |
Never run, or queued for the agent. |
running |
Backup in progress; the row shows reported progress and stage. |
success |
The run completed and created a snapshot. |
failed |
A required stage failed, the run was stopped, or it timed out. Check the stage details. |
warning |
A filesystem snapshot was created after zero files were processed. Check the source path. |
stalled |
A running job has not reported progress for many hours. |
force-run |
Queued for the next agent check-in, even if the job is disabled. |
Run now queues the job; it does not start a process immediately. Stop asks the agent to cancel an active run. Mark as failed closes the run in the dashboard and also requests a stop; use it if the agent is unreachable or Stop does not resolve the run.
Edit or delete a job
You can change the source type and settings, hooks, destination, schedule, and enabled state. Disabling a job prevents scheduled runs but does not prevent Run now.
Warning
Changing the destination requires setup again because destination credentials are keyed to the
destination. Changes to PostgreSQL connection settings may also require setup again. Run
sudo vecta-agent setup <JOB_ID> on the agent before the next backup.
Deleting a job permanently removes its run history and snapshot-ID records. Repository data is not deleted, but Vecta can no longer find those snapshots in the dashboard. See Restoring Data for the recovery path when job records are gone.
View a job's snapshots
From Jobs, choose Snapshots on a job to see the snapshot IDs reported by that job, newest first. The page shows run time, backup type, Restic counters, and whether Vecta considers each snapshot restorable. Use Restore to get a manual restore command. A Pruned snapshot is marked unavailable and has no restore action.
This is the history Vecta has recorded for that job, not a full inventory of the destination repository. To inspect repository snapshots that are not listed here, use Restic directly on the agent machine. See Restoring Data.
Notifications
A failed job sends one email to the account owner from noreply@notifications.vectaapp.com. It
includes the job name, server name, and a short agent error message. Vecta does not send success or
warning emails.
What's next
See Database Backups and Hooks for PostgreSQL and hook setup, Restoring Data to test recovery, or Job lifecycle & statuses for state transitions.